Foucault defines technologies of the self as practices by which individuals act on their minds, bodies, conduct, and way of being, to transform themselves toward some desired state. It’s an interesting lens to view modernity through, as it asks us the fundamental question: What are we capable of becoming?

Foucault’s ideas have always fascinated me. They have a strange way of enduring despite a world that changes around them. His work offers a value framework, a way to turn the idea of technology on the self. We so often view technology through the lens of metaphor - we never merely see; we always see as. In this sense, the internal gaze is misplaced in the direction of something or someone else as a way to explain it, or make it familiar to us. By repositioning the ‘gaze’ on ourselves, we are forced to confront the practises that make up the raw materials of our being. This is interesting for a few reasons.

First, he was explicit that these practices were techniques for working on your own thoughts, conduct, body, and soul, to transform yourself into a state of happiness, wisdom, or perfection. Technology, then, is just another way to improve ourselves. It’s how we improve that matters here - the process. Foucault saw this ‘improvement’ as an act of self-love. Social media didn’t exist whilst he was alive, but the idea of the ‘visible persona’ has always been present. Social media (Instagram, TikTok, LinkedIn) amplify the image of the self because for many, how we look has become who we are. This is the visible persona - the mirage we want people to see. There is a perception that chasing impossible perfection forces us to look within ourselves. This is the internal gaze. As a conceptual framework, Technologies of the self asks the question: why do we cultivate our minds and bodies, and who do we do this for?

As we continue to make our memories in visual ways, we become a product of that ideal. I have often wondered about the motivation behind the ‘selfie ’- is it vanity, or a modern necessity? Or, is it something closer to Foucault's care of the self: a daily, almost ritual act of composing who we are for an audience, curating a version of ourselves we'd like to become. Every selfie is proof of where we were, who we were with, who we're becoming. This, done with the intention of self-care, improvement or even transformation, speaks to our innate drive to be seen in a certain way rather as we really are. In this practice, our authentic self is different from our real self, as what people see is a projection of our desired ideals.

Foucault traced technologies of the self back to Christian confession — you become a legible, governable subject partly through the act of narrating yourself to an authority. In today’s world, we engage in the practices of improvement through digital transformation, or at least we tend to use digital technology to document it: fitness trackers, mood apps, AI chatbots you journal to and a smartwatch to record your improvements. This thinking extends further still. A child using AI to ‘improve’ an essay, or confirm their ideas, illustrates how they are doing technologies of the self in Foucault's precise sense - to improve themselves, whether that be product, knowledge or perceived outcome. In this way, we are training to improve ourselves based on something that is already there within us.

Foucault was careful to note that technologies of the self never operate alone; they sit beside technologies of power, which shape and direct individuals toward particular ends. The selfie, the tracker, the chatbot we journal to: each promises self-knowledge, but each is also quietly authored by someone else's design, someone else's metric of what "improved" looks like. So the question this way of thinking poses isn't only what are we capable of becoming - it's whose idea of becoming are we working towards?

The movie’s central metaphor, ‘Growing up and letting go,’ drives the narrative. It underpins the plot and points to one of life’s other eternal enigmas:

What does technology want?

This is not a new question, of course, and it’s one we have asked many times before. On first inspection, you might think it curious or even bizarre to assume technology wants anything. I first came across this line of thinking when reading Kevin Kelly’s book of the same name. If you think about it carefully, technology wants to be useful, to have a purpose, as do toys. Toys are just another form of ‘technology’- another thing we have invented in the image of us. That image is so often communicated via metaphor, as we never merely see; we always see as. Take AI for example; we often try to domesticate this technology to give it meaning - AI as a black box, AI as a mirror, AI as a benevolent genie, etc. Our relationship with metaphor is a complicated one, but our relationship with technology and the innovations it affords is less so.

As Toy Story 5 plays out, we come to see ‘toys’ as cultural objects that endure through our innate desire to play with them. They are, in essence, ‘objects’ that connect to our uniquely human histories. In this sense, the toys themselves embody knowledge and experience shared between generations. For example, the toy makers design the toys to capture the imagination of the children who play with them. The children who play with the toys create their own narratives through play. The toys in the movie (Woody, Buzz, Jesse, etc) want to be played with. This is what this particular technology wants. When a new technology comes along, the old technologies compete for attention and fear becoming obsolete. Like other Toy Story movies, old toys are often depicted as relics that lurk in the dark underbelly of the discarded, confined to history as they lose relevance and the world evolves around them.

Rarely has a movie written for kids captured my thinking in such a way. In a little over two hours they were glimpses of Kant’s ‘the mind interprets the world’ and Foucault’s Technologies of the self that left me to ponder the deeper philosophical questions of our age, like what are the truths behind the truths? Who are the objects, and who are the subjects of our gaze? In one scene, the toys fight to become the object of their child’s attention, evoking Judy Wajcman’s notion that technology can be seen in three ways:

• The physical objects themselves (the toys)

• The human activities that take place in conjunction with these physical objects (playing)

• The human knowledge that surrounds these activities - ‘what people know as well as what people do’.

In the movie, the protagonist is a young girl who is shamed for playing with toys by her peers who have shifted their attention to the lure digital screens. As they tap in a zombie-like state with TikTok attention spans, she feels isolated, which speaks to one of technology's great ironies - how devices designed to unite us can serve to isolate us further from what we crave - human connection. In this sense, the movie felt a decade late - perhaps the kids should have personal AI’s or robots, but it worked, and the message landed. I wondered how many parents in the audience felt a pang of guilt and self-reflection as the infectious nature of technology took over numerous households on the screen.

As the movie reached its conclusion, I was left to ponder the true nature of technology. That is, the fundamental, existential question: What is technology?

In basic terms, technology is understood as the process by which humans modify nature to meet their needs and wants. The academics Lievrouw & Livingstone (2002) framed the interconnected nature of technology, and our relationship with it, through three critical domains:

Artefacts and devices: the technology itself as a physical object

Activities and practices: what people choose to do with technologies.

Context: the social arrangements and organisational norms that surround the use of technology.

I like this framing, and it works well with what was unfolding on the screen as it perfectly captures the human and the non-human potential of technology to reach us on a profoundly personal level. This, I suppose, is a quality of our own consciousness, as my experience differed greatly from my 5-year-old granddaughter's despite watching the same thing, at the same time, in the same place. I am fairly certain she saw beauty where I did not, and I rather doubt Kant or Foucault were occupying her thoughts. Despite these obvious differences and the age gap, we both enjoyed the film.

As the credits rolled and people began to leave, a thought washed over me. Life, and the moments that define it, happen when we dare to use our technologies in new ways, or even when we leave them behind and choose to look away entirely.

Deep into the credits, another truth revealed itself, one that was lost on my granddaughter. No spoilers here, so you’ll have to watch the movie until the very end, but if you are a fan of another space-related Sci-Fi franchise, you’ll know the line ‘I am your father’. That line gets put to work here.

Nice touch by the directors.

Last year, I set about trying to understand this divide, so I embarked on a systematic literature review as part of my Doctorate. The full 8k words are accessible here: Gender Participation in STEM Education: A Systematic Literature Review of UK Secondary School Contexts to identify equity interventions.

Here is the 800-word version.

In short, there are three entrenched, and at times intertwined, reasons why Girls’ participation in STEM subjects reflects structural conditions that systematically limit their collective recognition as legitimate STEM people. Let’s break it down.

1. Formation of STEM Identity

The research spotlights how early STEM experiences (primary, foundational years) shape a sense of belonging. Where children are exposed to STEM activities in a neutral setting (both genders have the same access to activities and learning), then girls thrive and are statistically more likely to study STEM subjects at GCSE, A-level and University level. But it’s a complicated story. STEM identity is not only shaped by school experiences, but also by how STEM roles are viewed and talked about in the home. This is where the ‘entrenched’ views can derail the most ambitious girls and force them to question whether they are ‘STEM people’.

My reading taught me that participation is driven by whether students see themselves as legitimate “STEM people”, not just by attainment. The research illustrates that STEM identity is shaped by three factors: early experiences (play, exposure, media), recognition from teachers, peers, and family and access to STEM-related opportunities. Girls often perform well in STEM subjects, enjoy STEM lessons, but still do not identify with STEM careers. That’s the ravine I was referring to in the opening paragraphs. In short, STEM identity is granted through recognition, not simply developed individually.

2. Cultural reproduction of gender inequality

This was an interesting, but not a surprising finding. I remember being struck by the words of Maths professor Jo Boaler (author of Elephant in the Classroom), when she describes a scene where a young girl comes home after a challenging maths lesson and says to her mother, ‘I found maths really hard today, mum.’ The response still flaws me today - Mother to daughter … ‘Don’t worry darling, I wasn’t any good at maths either at school.’ This is a classic example of stereotype, or the entrenched threat of intergenerational, gendered beliefs. The mother normalises struggle as a permanent, inherited trait rather than a developmental challenge.

Cultural beliefs are shaped at school, too and far too often the gender disparities above are systematically reproduced within our places of learning, which are by design, supposed to be neutral - our curriculum, our pedagogy and our expectations. The research points to well-intended interventions, focused on ‘fixing girls’ through raising the profile of STEM subjects, and their role within them. There are intentional contentions, too, such as the language that surrounds STEM, which consistently frames it as masculine. This impacts subject choices and eventually influences STEM career aspirations. One is symptomatic of the other. We know that in the UK, where this paper is grounded, girls drop out at key transition points (GCSE, post-16), with the ratio of boys who study STEM subjects far higher. This is true even when girls outperform boys in science, maths and computer science from an early age. I saw this historical trend play out in the research.

Away from schools, gendered norms are reinforced at home, with numerous studies suggesting parents actively discourage girls from studying STEM subjects. I worry that our obsession with putting female role models in front of girls. There is much to be said for the benefits of this - representation in the labour markets, career progression and a duty of care, but the data is stark. In almost every case, the impact of the visiting STEM expert dwindles once girls return to their norms - whether that be in STEM lessons or at home. One solution might be to fix our gaze on making STEM subjects inclusive rather than changing structural conditions. The message in the literature is clear - inequality is reproduced by cultural norms and systems, not individual deficits.

3. Intersectionality of gender and socio-economic disadvantage

You might not be surprised to learn that gender inequality in STEM is not experienced equally. It is intensified by socio-economic status. The research shows us that girls from low-income backgrounds face compounded barriers. This appears in the form of limited access to STEM capital (resources, networks, experiences) and less support from families for STEM pathways.

STEM capital is a constant in the research surrounding gendered experiences of STEM education. The term ‘capital’ was popularised by Pierre Bourdieu (1972) and found its way into education literature through various forms - cultural capital being the most popular. Louise Archer and colleagues developed science capital around 2013 to 2015 by applying Bourdieu's framework specifically to science education. Science capital refers to the science-related knowledge, attitudes, experiences, social contacts, and resources that make it easier for some young people to see themselves as ‘science people.’ There are links here to the writing above. The lesson here is that gender inequity in STEM is amplified by wider social inequalities, not just gender alone. Participation gaps appear to be structural, not ability-based.

But there are some exceptions, so let’s look at the outliers.

Girl-focused STEM interventions:

Many interventions (e.g. enrichment, clubs) fail to close the gap long-term, especially if access is unequal or funding is inconsistent. In many cases, the funding dries up, the club closes, and the opportunity falls away. Or, the teacher driving the innovation leaves the school, and nobody replaces them. Despite this, the outlier data shows some snippets of hope. When STEM clubs involve both genders, and girls are treated equally to boys, they thrive. In one study, where girls were asked if they would like to lead STEM teams in a Formula 1 task, the researchers reported higher levels of confidence and belief from the girls involved.

Misdiagnosing the problem leads to ineffective solutions.

Interventions are not solutions; they are sticking plasters applied to a wound that requires surgery. This wound is the entrenched view that STEM subjects are not for girls. If you are reading this and you are the exceptions, I would be fascinated to hear about your journey. The reading taught me one painful lesson - we have spent decades fixing girls for a system that was never broken for boys. Changing the nature of STEM activities to make them more inclusive for girls has worked in a few cases, and I spotlight this in the full version of the paper here: Gender Participation in STEM Education

Short-term solutions to an entrenched problem

Many interventions are simply not given time to breathe. They are either short-term or lack longitudinal impact. The main barriers to these interventions scaling is that they rely on funding that has a shelf life, or they need specialist staff to run. The problem with this is twofold. Firstly, it serves to reinforce existing inequality when the opportunity is taken away, or access is limited. The other more sinister issue is the framing of girls as the problem that we need a solution for. Effective change requires system-wide transformation, not isolated initiatives.

Drawing conclusions that might spark future solutions

Gender inequality in STEM is not an individual problem; rather, it is a system that works exactly as designed. Schools, families, and labour markets do not operate in isolation; together, they form a self-replicating cycle that normalises who belongs in STEM and who does not.

Interventions that focus narrowly on girls’ confidence or aspiration, without questioning how STEM itself is framed, risk reinforcing the very inequalities they set out to challenge. Real change requires disrupting the cultural logic that positions STEM as masculine by design, or rethinking how STEM careers are represented to young people. This forces us to confront the inequities that persist even when girls succeed academically. Without that systemic shift, gender inequality in STEM will continue to reproduce itself quietly, persistently, and across generations regardless of rising attainment or renewed policy ambition.

Thank you for reading this article. The words are the authors’ and no Generative AI was used during the writing of the words above. For full access to the paper and for an extensive bibliography, please click the link here: Gender Participation in STEM Education: A Systematic Literature Review of UK Secondary School Contexts to identify equity interventions.

Scopus overview

This opens the dialogue around what oracy and AI can do for education. At first glance, you might think this is a curious mix, but taken together, these apparently separate fields offer us an array of ‘teachable moments’ on advocacy, agency, and adopting a critical stance in an AI-mediated world. This is why I floated the idea of an AI through Oracy series as a collaboration between Voice 21 and STEM Learning. Children don’t just need to understand AI. They need to know where they stand on it. Bringing this moment to life through talk will teach them to build their own stories and opinions about this technology. Oracy is the vehicle for this; it gives children the language and confidence to form, test, and own their views. This is where you come in, the teacher. Your role in this is to create the conditions for speaking and listening so AI and oracy have space to breathe in the curriculum.

The importance of oracy in the AI debate

At the Voice 21 Speaking Summit last month, language was spoken about as the embodiment of thought. The Oracy Framework, developed by Voice 21 and Oracy Cambridge, gives teachers a way to define speaking, listening and communication skills across physical, linguistic, cognitive, and social-emotional strands. By contrast, AI has been on its own linguistic journey, one shaped by large language models (LLMs). Where oracy developed through human relationship and pedagogy, AI developed through pattern, prediction, and scale and yet both now sit in the same classroom, asking the same fundamental question: what does it mean to communicate? This is why AI needs to be talked about, and why oracy as a medium to share ideas matters.

Some teachers would rather turn down the volume on AI, but it would be irresponsible of us to pass over it in silence. If we are alive to the possibilities of what AI can offer our students, then we open the door to dialogue. But when it comes to AI, it is more than speaking or being heard; it’s about listening to what students think and speak. Listening, then, is an ‘orientation’ or a way of positioning the talk so it matters. What this series of booklets attempts to do is empower teachers to speak to children about AI. If we can achieve that, oracy gives children the language to recognise the inequalities AI can quietly embed and the confidence to push back. Whether this ‘pushing back’ is in relation to deep fakes or AI’s impact on the planet, allowing students to be active in the debate by taking a stance builds agency. The ideas and activities in this pack have been designed with the child in mind, with provocations, probing questions and the surfacing of the language that surrounds AI. If oracy is the embodiment of thought, then these activities are where that thinking begins.

How to approach this series of booklets

This three-booklet series provides teachers with structured, progressive lessons designed for a range of teaching moments that work as standalone activities or integrate into the wider digital curriculum. Each booklet can be taught within PSHE, tutor times, or other appropriate moments and returns to one essential question with increasing sophistication across key stages. Teachers receive explicit learning objectives and vocabulary definitions to help navigate the content. The booklets have been written with age and stage in mind. There are three distinct access points to the content: EYFS & Key Stage 1, Key Stage 2 and Key Stage 3, covering learning to talk to learning through talk.

Booklet 1: What is Real? Download now. We follow one question from a four-year-old spotting a shark on a discovery table to a fourteen-year-old debating deepfake responsibility. In KS1, children learn to question what they see. By KS2, they are spotting manipulation. At KS3: not just can I spot a fake, but what should we do about it?

Booklet 2: Can We Trust It? We discuss cartoons, talk tactics, and debates to establish that truth is rarely straightforward and trust must be earned. In KS1, a story told two ways starts the conversation. By KS3, the question is systemic: who controls the information, and what happens when trust breaks down at scale?

Booklet 3: What is the Impact of AI on the Planet? The final booklet explores land, water, energy, and minerals. In KS1, children learn that devices are made, powered, and discarded. By KS3, the question becomes why and who, examining why environmental costs fall unevenly across the supply chain.

Together, oracy and AI offer education something it has always needed but rarely had a clear language for: a way to help young people make sense of the world they are actually living in. These booklets are not a curriculum solution or a policy response; they are an invitation. An invitation for teachers to create the conditions where children can think out loud about technology that is already shaping their lives, form opinions that are genuinely their own, and develop the confidence to voice them. That invitation extends across the curriculum.

The questions these activities raise don’t belong to a single subject; they touch the boundaries of many. Like literacy, oracy is often left to English teachers - despite being essential across subject domains. Conversely, AI tends to sit within Computer Science, which isolates it from other subjects. In truth, the issues within these booklets sit at the intersection of science and society, of data and democracy. When students explore the environmental cost of AI, they are doing geography and ethics. When they interrogate a deepfake or trace the origins of a viral claim, they are practising the same critical scrutiny that underpins good science, responsible citizenship, and media literacy, threads that run through the 2025 CAR. I would argue that teaching AI through talk is not an abstract concept, but the connective tissue of a cohesive curriculum, one that prepares children for an uncertain future. Teachers across STEM, humanities, and beyond will find natural entry points here, and that cross-curricular dialogue is part of the point.

If we get this right, the young people who move through these experiences will leave with more than just an understanding of AI. They will carry a habit of mind to question, to challenge, and to think critically about the systems around them. That matters because the future won’t just happen to them; they will shape it. And perhaps more importantly, they will begin to understand their place within it, not as passive users of technology, but as active, informed, and thoughtful participants in a world they have every right to interrogate.

For those who teach, I hope you enjoy the activities. For those with a keen interest in promoting free speech and empowering young people to communicate their ideas, if you know any teachers, please pass this resource on. Here it is: What is Real? Booklet 1

Booklet 2 drops on May 20th and Booklet 3 on April 15th.

When it comes to the delicate matter of impact, AI treads violently. As technologies go, this one is heavy going. Born in the muscular legacy of capitalism, AI companies want us to internalise one form of progress at the expense of another. This throws up an array of promising contradictions, and as consumers of AI products, we often find ourselves trapped in the middle.

All the problems are all the problems

Framing the impact of AI on the planet is complicated. For this article, I have borrowed from the work of Hlabisa (2025) and used four critical domains which categorise the environmental impact of AI. They are:

1. Energy

2. Water

3. Materials

4. Land

What follows is a brief exploration into each domain, casting a critical eye on the AI companies that are driving to scale this technology at the expense of digital convenience, and of course, progress.

Energy

Large Language Models (LLMs) require vast amounts of water to train. Global data centre electricity consumption is estimated at around 415 TWh in 2024, roughly 1.5% of global electricity use, having grown at 12% per year over the last five years. About 60% of a data centre's electricity powers the servers themselves. These are AI-optimised hyper-scale facilities that use advanced chips capable of trillions of calculations per second, which require two to four times the wattage of conventional chips.

Cooling systems account for the next largest share, anywhere from 7% at efficient hyper-scalers to over 30% at less efficient facilities. This is where our next critical domain enters the cycle - water.

Water

Data centres generate heat as the TPUs (chips) generate heat, which needs cooling. Water plays a dual role in the AI production chain as it is used for cooling and generating electricity to power the servers.

Cooling

When you put huge amounts of electricity through a heavily concentrated bank of servers, the heat is intense. If left, it would literally melt the silicon. To counter this, massive evaporative cooling systems are used to moderate the temperature of the servers and keep them optimal. In this setting, water is either consumptive (removed from the source but never returned) or non-consumptive (used to generate the electricity that powers the data centres). This matters too, as in both scenarios, water is not returned to its source, which leads to scarcity, and in some places, worsens droughts.

The science is worth sharing here. When water evaporates, it goes through a phase change from liquid to gas. This process absorbs heat. Data centres pump hot air from the servers (cooling) and spray freshwater onto the source of the heat (servers again) via cooling towers. Water evaporates, pulls heat away and escapes into the atmosphere. The average data centre consumes 7,000 to 12,000 litres of water for every hour of use, and they stay open 24 hours a day, 365 days a year. This is a problem as the servers are thirsty and consume vast amounts of water from local freshwater sources. But the story doesn’t end there. We also need to consider water use in the production of the chips in the first instance.

The physics of computing dictates that only ultra-pure water can be used when producing chips. Transistors are manufactured at the nanometre scale, and even microscopic impurities would disrupt the fabrication process and damage the circuits. No dust or minerals can touch the surface of the transistors. Through reverse osmosis, deionisation and UV Light, every strip of water impurity is stripped until water is reduced to its bare H20 molecules. This is a battle as water doesn’t like to be pure, as it becomes a solvent and wants to absorb minerals.

Materials

When we talk about AI, the conversation usually focuses on software, algorithms, and data. But AI is ultimately physical. Behind every model and every prompt sits a vast global infrastructure built from minerals, metals, and rare earth elements extracted from the planet. This extraction is to meet demand. This is best described by Jevons ’ Paradox, where efficiency improvements ultimately lead to greater overall resource consumption. Jevon, an Economist from the 1800’s, noticed that when Steam Engine efficiency increased, coal consumption increased as more people travelled, more often. And the cost is reduced. The same ‘paradox’ applies to technologies, and specifically LLMs. In the modern day, this leads to what Hlabisa (2025) calls Anthropic Escalation.

To understand the tension here, it helps to focus on the AI supply chain. Minerals used to create the physical compute components are mined in Africa. The chips are fabricated in China. The LLMs (models) are trained in the USA, moderated for abusive and dangerous content in Rwanda, and then sold for consumption globally. This infrastructure is globally distributed, yet the impacts are heavily localised and native to the people who occupy that geographical location. These processes are often invisible to the consumer and are well hidden by the companies that commission them. AI infrastructure also depends on several critical minerals, including: Gallium, Germanium, Indium, Cobalt, Lithium and Tantalum.

Many are mined in geographically concentrated regions. For example:

• Cobalt production is heavily concentrated in the Democratic Republic of Congo.

• Rare earth processing is dominated by China.

• Advanced semiconductor fabrication is largely carried out in Taiwan and South Korea.

This concentration introduces both environmental pressure and geopolitical vulnerability. The global AI ecosystem depends on a supply chain that stretches from mining sites to land to house data centres, which brings us to our final critical domain - land.

Note about GPUs and TPUS: At the heart of every data centre is a hierarchy of specialised silicon. The workhorse of AI training is the GPU (Graphics Processing Unit). Alongside GPUs sit CPUs (Central Processing Units), which manage the flow of instructions and data to the GPUs.

Land

The modern data centres have a footprint of 30-100 hectares. That’s the equivalent of 56 to 187 football pitches. It’s vast. If you add the concrete foundations, the physical data servers, the structures and electrical sub-stations, it’s easy to see why residents are rising up against data centres being built in their towns and open green spaces. This conversion from green spaces, or wasteland, to industrial zones is creating new problems, such as chemical pollution, habitat loss and water scarcity. As AI models scale and the race for AGI (Artificial General Intelligence) accelerates, so do the land grab tactics. Ireland and Singapore are ground zero in this race to acquire land. Data centres in these locations are crashing the local grids, leading to blackouts and energy loss. This compounds an already fragile problem.

Inventive entrepreneurs are looking to the deserts of Dubai and space as a solution. Dubai offers vast land, cheap labour costs, and an abundance of solar energy, maybe too much. Solar panels overheat under intense radiation, so strong sunlight means large amounts of energy hitting the panel. As the panel temperature rises, electrical resistance increases, voltage drops and efficiency decreases. So it’s not the ideal solution. Housing a data centre in space is sometimes proposed as a future idea because space offers a few physical advantages for computing infrastructure. If you are Elon, and happen to own a rocket company, have an AI model and satellites - this could be an attractive, rent-free solution. Putting computing infrastructure in orbit would avoid land use, zoning conflicts and water consumption for cooling, as Space is extremely cold (around –270°C background temperature), so heat can be removed through radiative cooling.

Future solutions

As AI systems scale and the race to AGI heightens, the demand for critical resources will only increase. Criticality must become the central dialogue for how we progress in a future that seems to be mapped out for us. Yet critique, however necessary, is not enough on its own. Paulo Freire called critical hope: the refusal to accept that what exists is all that is possible. If AI treads violently, the question becomes who bears that weight, where, and whether those of us who benefit from it are willing to look.

One solution that holds promise is follow-the-wind computing. This process dynamically shifts workloads to wherever wind energy is currently available, with related strategies like co-locating data centres directly with offshore turbines, and using curtailed or ‘stranded’ renewable energy that grids can't absorb. It would make compute cheaper and greener in ways that could cope with demand (remember Jevons' Paradox). The geography of wind resources favours the Global North, shifting the burden on the Global South. Maybe the future of data centres is not deserts and space, but oceans and the vast open spaces.

Critical hope demands something harder than technical innovation. It asks us to recognise that the future is not, as AI companies insist, a singular path ahead. It is contested terrain. The question of who bears the weight of progress, and who gets to define what progress means, is not a technical question; it’s a political one. And it will not be answered by where we put our data centres. It will be answered by whether we are willing to look at who is paying for them.

References

Hlabisa, S. (2025) ‘The ecology of artificial intelligence: energy, water, materials, and land limits of digital systems’, Carbon Neutral Systems, 1, p. 19. Available at: https://doi.org/10.1007/s44438-025-00018-8 (Accessed: 12 March 2026).

Freire, P. (1994). Pedagogy of Hope: Reliving Pedagogy of the Oppressed. New York: Continuum.

If you are curious what an AI Sprint is, I will do my best to define it. In essence, it’s a fast-paced exploration into the issues and ethics that surround artificial intelligence (AI). It’s not a podcast as it’s a live event. It’s also not a webinar or a workshop. Instead, it’s a live community event that opens the door to critical dialogue, ideas and a healthy dose of honesty about a technology that is evolving faster than many of us would like. The narrative is grounded in the school and centres as it should be, around the ‘child’. As we approach the end of Season 2, I feel proud of what it has become.

AI Sprints has an origin story

Picture this: an online STEM Community event about AI, with hundreds of people tuned in from all corners of the world. The energy was electric. For a full hour, we explored the world of artificial intelligence together, and the audience was captivated. Nobody left. By the time we wrapped up, the comment section had exploded with over 800 messages. People were curious, excited, and hungry for more. Questions were flying, ideas were bouncing around, and it became crystal clear that this conversation was just getting started. That’s when the AI Sprints were born. What began as a single live event took on a life of its own.

I am deeply fascinated by how ideas that are shared in one setting find new life in another. The live sprints have become a testing ground for ideas and a space to share disruptive and at times, where money is scarce, frugal innovation.

As with many live experiences, the guests shape the narrative. My role as host is to find the authentic people and help share their insights. I am part journalist, part script writer, and then there is the delicate matter of the questions – how to provoke just enough in the right places, at the right time. I try to ask questions with just enough bait for the ideas to reveal themselves, then I nod, wait and smile, and let silence do the heavy lifting. From my years working as both a teacher and a researcher, I have learned that the truth frequently reveals itself not in words, but in the spaces between them. I refer to these as the silences. Hosting is equal parts therapy, trust and the occasional trick. Some days it flows. Other days, you must work a little harder and prompt in the right places.

Finding the reliable voices in education

The methodology for the Sprints was to seek out the most reliable voices. These people are often found on the fringes of education, acting alone. In a profession so resistant to change, innovation can often be seen as an act of defiance. I see it as an act of love. I have always admired those folks who have the courage to disrupt from the inside out. They have mastered the art of being agile within the box. Conversely, there are those who work in groups or communities, and they are interesting too.

When I reflect on the episodes to date, it fills me with hope. Education is full of people who care deeply. In this sense, the AI Sprints beautifully enacts its own thesis. Both what we notice and how we think about noticing matter.

Reflecting on the journey so far

Communities of practice can be transformative in education. I think as humans we are conditioned to be critical, it’s our default setting. Opening space to hear new voices can challenge our worldview and expand our knowledge beyond our familiar ‘echo chambers’. I have learned so much from the guests I have spent time with. For example, Laura Knight’s poignant assurance that we need to ‘think with care’ when it comes to AI has stuck with me. There are others, too. Too often, those at the forefront of AI speak with perfect certainty. However, it’s refreshing to hear from people who don’t know. This has given me an unshakable conviction that education values truth, and often we find it in the most unexpected places.

AI Sprints episodes to date have shown an appetite for alternative conversations about this technology. For most of the sprints, other aspects of teaching were the central focus. Darren Coxon, Priya Lakhani and Dr Kate Pardine championed ‘voice’ as a transformational force for good in the classroom. Teachers should be speaking to children more. Lord Jim Knight and Daisy Christodolou spoke about knowledge, and Bukky Yusuf and Al Kingsley showed us how bias, a flaw as old as time itself, is ever-present in everything we do. These are very ‘human problems’, exemplified by AI rather than problems defined or created by it.

BETT as a testing ground

Last week, STEM Learning took a brave step by bringing the AI Sprints to BETT. I felt extra pressure as it was my idea. I wanted to test whether the vibrant online community could work in a live in-person setting. BrainFreeze, long-time sponsors of the Sprints, kindly offered to host us on their stand. The setting gave us a presence at BETT to share more stories. In the weeks leading up to the event, I worried if anyone would show up to listen. But they did show up. Not just a handful of curious passersby, but people who were there to see it live. At times, it was standing room only, with people pressed shoulder to shoulder, leaning in to catch every word. And here’s what struck me most: very few people left. They stayed, engaged, asking questions, sharing their own experiences. At one session, a woman stood beside another attendee, gracefully signing throughout the entire presentation to ensure that everyone could be part of the conversation. It was a beautiful reminder of what inclusive engagement truly looks like, and it made our captive audience even richer for it.

I feel incredibly privileged that so many people chose to give us 30 minutes of their time at BETT, a show where every minute counts and there’s always somewhere else to be, something else to see. That they stopped, stayed, and became part of our conversation means everything. I was inspired by the support the education community gave us at a time when there was so much noise and uncertainty. It was also great to welcome back previous guests and meet so many people in person, people who have become fierce and loyal Sprinters themselves.

A final conviction that proved that AI Sprints can exist in new formats came when Trudi Barrow and Aaron Patching took the stage on day 2 of the conference. Rather than express in words the gratitude people felt when this session ended, I thought I would share the image below. Theirs is a story people wanted to hear. A live walkthrough of some of the most creative ways that teachers are using AI to enhance students’ work. Pure brilliance!

The AI Sprints have proven that authentic conversations about AI in education matter. When we create space for honest dialogue, grounded in real classroom experiences and centred on children, people show up. They listen. They stay. And together, we find the stories worth telling, the ones that transform education from the inside out.

How to join future AI Sprints

If you would like to join in future episodes of AI Sprints, simply scan the QR code below or sign up via this link for free: https://forms.office.com/e/LSgh5MBUH3

A number of people have since asked to hear more about the thinking behind it, so this piece is the first in a short three-part exploration of the concept.

Image by DALL-E (2025)

The image above illustrates how ChatGPT and CoPilot view the AI Epoch as a ‘rupture’ or ‘event’- interesting!

‘History, viewed on the largest scale, appears to unfold through a sequence of distinct growth modes, each vastly more rapid than the one before’

Nick Bostrom from the Future of Humanity Institute at Oxford (Bostrom, 2014).

The AI Epoch

Epochs represent an epistemic shift as they emerge as underlying structures that define what counts as valid knowledge in a given historical period (Foucault, 1966). In an age of AI, the human becomes both the ‘knower’ in the form of the subject of inquiry and the ‘known’ as the object being studied. AI fundamentally transforms the epistemological landscape by creating a reflexive dynamic where humans simultaneously occupy dual positions in knowledge production. This epochal shift redefines scientific practice by expanding what constitutes legitimate knowledge and methodology.

To interrogate this moment, Foucault’s epistemes and Freire’s epochal units offer essential tools for naming the AI Epoch, revealing how AI-systems recast the very conditions in which knowledge, power, and agency take shape. It is these three variables: knowledge, power and agency that are defining characteristics of the AI Epoch.

Foucault’s work on epistemes shares many parallels with the AI Epoch - noticeably through the defining characteristics of how breaks in knowledge are structured. These breaks, described as ruptures by Foucault (1966), are linear progressions, each with rules for what counts as knowledge. In contrast to, yet also in support of, the notion of historical ruptures, Freire’s concept of epochal units frames social transformation as distinct historical periods marked by particular themes, contradictions and social struggles that shape the consciousness and material reality of the people living within them. Freire’s view differs from Foucault's in that he believed epochal units should be defined by the generative themes of the time. This encompasses the dominant ideas, power structures and the tensions that characterise life.

If I had to define the AI Epoch, it would be as follows.

The AI Epoch is a proposed historical period defined by the pervasive integration of artificial intelligence into social, economic, cultural, and educational systems.

The problem with defining something that is fluid is that it shifts and takes many forms, so it proves hard to domesticate. Rather than trying to define it, it makes sense to place some boundaries around it and see where they go.

The 5 Domains of the AI Epoch

‘We are the machine, so we are responsible for the boundaries’

Donna Harraway, Cyborg Manifesto (1985)

The AI Epoch is not emerging but present, and appears as a discontinuity in five key domains. These are:

Ontological: AI challenges fundamental notions of what constitutes intelligence, consciousness, and human uniqueness, forcing us to reconsider the nature of being and whether machines can possess forms of existence previously reserved for humans.

Epistemological: AI disrupts traditional knowledge creation and creates new dynamics between the human ‘knower’ and the artificial ‘knower’ (the AI) to govern what counts as truth in an era of generative text, code, audio, video and image.

Economic: AI begins to reshape labour markets and value creation, potentially displacing workers while concentrating wealth among those who control AI systems, fundamentally altering the relationship between work, productivity, and economic participation.

Social: AI reconfigures human relationships and social structures through algorithmic mediation of interactions, surveillance capabilities, and the emergence of ‘synthetic interactions’ with non-human entities. This movement exists within the existing field of Posthumanism

Pedagogical: AI transforms teaching and learning by challenging traditional notions of skill development, assessment validity, and the teacher’s role, thereby disrupting existing power dynamics associated with knowledge.

In part 2, I will interrogate the five domains to explain why I feel the AI Epoch is reshaping knowledge, power and agency.

References

Bostrom, N. (2014). Superintelligence: Paths, Dangers, Strategies. Oxford: Oxford University Press

Foucault, M. (1966) Les Mots et les choses: Une archéologie des sciences humaines. Paris: Gallimard. (English translation: The Order of Things, 1970, London: Tavistock.)

Freire, P. (1970) Pedagogy of the Oppressed. New York: Continuum.

Haraway, D. (1985) ‘A Cyborg Manifesto: Science, technology, and socialist-feminism in the late twentieth century’, in Simians, Cyborgs and Women: The Reinvention of Nature. New York: Routledge, pp. 149–181.

More, A. (2025) The AI Epoch. Presentation delivered to the AIEOU Community, University of Oxford, March 2025.

More, A. (2026, forthcoming) ‘The AI Epoch’. In: Ethical AI and Data Science. Taylor & Francis

UNESCO (2020) Artificial Intelligence and Education: Guidance for Policy-makers. Paris: UNESCO.

Some light reading

You might think it curious that I picked this topic to break a two-month silence on the blog. The truth is, the ‘posthuman’ movement has been bothering me for a while, ever since a theme surfaced in my doctoral research. Starting in 2002, I began to interview students aged 11-15 years about Artificial Intelligence (AI). I was interested in exploring student voice in relation to this disruptive technology when something strange happened. The children kept referring to AI as ‘they’. Spooky. But what did they mean?

What is posthumanism?

Rosie Braidotti, a pioneer in the field defines it as “a critical framework that questions human exceptionalism by decentering the human in modern thought and examining the interconnectedness of human, non-human, and technological elements.”

If we unpick this, some interesting themes and implications are lurking beneath the words.

As a critical framework, Posthumanism challenges the traditional notion of the human as exceptional. It argues that humans are not the sole measure of value, knowledge, or consciousness, and that other forces are at play. These forces are ‘non-human’ and might appear in the form of technology, AI, biology, and ecology.

As a field of research, it takes its roots from established ways of thinking about the world. Humanism is the obvious starting point - a position that places Humans as the dominant and moral centres of the universe (Descartes, Kant). There are also traces of Postmodernism, specifically in relation to metaphor, narratives and how we see ourselves in these technologies (see Technologies of the Self by Foucault). AI and cybernetics offer the promise of human enhancement, merging biology and synthetic entities to form an entanglement of sorts. Above all, the work of Donna Harraway and Rosie Braidotti, driven by feminism, dominates the posthuman landscape - hinging around hierarchies, gender, race and identity. But, despite these strong foundations, I have found it to be a messy, fragmented, and often self-contradictory field. It has intrigued me and frustrated me in equal measure. It refuses to be one thing. Instead, it is an entanglement of many things, living and otherwise, and this confuses me. I am not alone.

My struggles with posthumanism

Before I spiral into what I don’t like about it, there are some positives.

In an age where humans have altered the planet’s climate and ecosystems, posthumanism calls for rethinking ethics, beyond the human, toward planetary responsibility. It rejects the idea of a detached ‘mind’, insistng that knowledge, thought, and identity are always embodied, material, and interconnected with environments and technologies. I like this, and such a mindset shift would do wonders for how we treat the planet and ourselves. However, there is a darker side to it. In no particular order, here are my issues with it:

Competing epistemologies

Conceptually, I feel there is a vagueness to posthumanism. It is really hard to categorise as there are so many versions of it. There is critical posthumanism, philosophical posthumanism, transhumanism, and new materialism, and they don’t all agree. Some place the human as the centre of everything, others seek to ‘dehumanise’ us. These elastic definitions make it messy and hard to categorise.

Speculative Futures

Our obsession with defining what comes next encourages fields like posthumanism to exist. Perhaps we need it to separate ourselves from other living entities and accept once and for all that we can’t predict the future. Speculating what might happen is a profitable business, so this is unlikely to happen anytime soon.

Colonial Bias

Like most emergent technological movements, posthumanism suffers from inherent bias. In this sense, posthumanism is intellectually bold, but dangerously fragile. History favours the winners, and post-human ways of thinking carry Eurocentric assumptions. Many non-Western philosophies (Ubuntu, Confucianism, Indigenous worldviews) already valued interconnectedness long before posthumanists claimed it as their own territory.

Full of contradictions

This is a field that claims to reject human exceptionalism but still uses human frameworks (ethics, logic, reason) to argue its points. It leans on human language and borrows words from biology and the natural sciences to suggest our future will be an entanglement with non-living matter, a symbiosis of sorts. In this sense, it’s a movement that can’t escape what it tries to transcend.

Tech Bias

Posthumanism is overly theoretical to the point it starts to drift into sci-fi at times. It is also a challenging read. Much of the writing (especially in the humanities) is steeped in dense language from Deleuze, Haraway, and Braidotti, high on abstraction, and speculative fictions low on practicality. Assuming everyone has equal access to a technological future also ignores the fact that many don’t, and won’t. This is a contentious point. In celebrating ‘hybridity’ and ‘fluid identities, ’ it often ignores real-world structural issues like poverty, racism, and global inequality. And the fact that the planet is heating up.

In summary

While posthumanism offers valuable perspectives on rethinking our relationship with the planet and technology, it remains frustratingly fragmented and disconnected from practical concerns facing most of humanity. I gave it a good go. I estimate some 50 hours I will never reclaim. It taught me one thing, in trying to go beyond the human, posthumanism ends up face-to-face with the strange reflection of what it just erased.

Alex

I see this everywhere I go, and it’s not unique to AI adoption. The practice of ‘borrowing ideas’ in education is a time-honoured tradition. From behaviour cards to curriculum maps, schools lift ideas from one setting to give them new life in the next. However, there is a problem with this - an idea borrowed often loses its relevance when enacted in a new setting, as the people who created that solution did so to address an issue specific to their school and setting. The same applies to the practices and pedagogy surrounding AI.

What is isomorphism?

Isomorphism explains why spatially separate schools might exhibit, in different ways, very similar practices (Puttick, 2017). Since being on the road delivering AI training to schools and MATS, I have noticed an almost tribal use of AI language, tools and wrapper Apps. This extends to the very language educators use and the narratives that shape how it’s perceived.

The thinking around isomorphism can be mapped back to a famous paper by DiMaggio and Powell from 1983. They noted that when new fields emerge, they are often characterised by diversity. When the field becomes more established, there is an inevitable push to homogenisation. In education, In education homogenisation happens when schools adopt the same policies, teaching methods, or technologies. When imitation takes hold, curriculum content or assessment approaches become standardised across different contexts, so local differences in practice fade because everyone follows the same models or guidance. Through a mix of policy mandates, the urge to copy perceived leaders, and shared professional training, AI adoption in schools follows the same path.

Homogensing forces

Educational research happens through ‘organisational fields’ (Puttick, 2017). In the AI context, this includes everyone who contributes to the field - EdTech suppliers, AI creators, policy-makers, AI consultants and influencers. Each plays a role in forming the ‘field’ that surrounds AI, but the process of formation is more nuanced when schools become involved. To examine this closer, we need to make sense of the homogenising forces that cause schools to copy each other.

Time - Schools are time poor. With limited time and capacity, schools frequently adopt ready-made ideas from peers rather than design approaches from scratch. They do this to save time on design and testing, reduce uncertainty about outcomes, demonstrate alignment with credible ‘best practice,’ and meet immediate demands without overloading staff.

Lack of AI expertise - A lack of AI expertise in schools drives isomorphism as leaders and teachers, lacking confidence to make independent decisions, tend to rely on external templates, copy perceived leaders, and follow professional norms that promote similar frameworks and tools. This practice sees us copy the ‘visible winners’ to adopt AI at pace.

Weak Leadership - Copying other schools verbatim, without adapting to the local context or critically evaluating fit, can be seen as a sign of weak leadership. It avoids the strategic thinking needed to analyse a school’s unique needs, strengths, and constraints, and it signals a lack of vision by following whatever is already being done elsewhere rather than setting a clear direction. This approach also risks a poor fit, as what works in one school may fail in another due to differences in culture, resources, or student demographics.

The ‘forces’ above show how easy it is to get AI adoption wrong. Imitation alone does not foster creative solutions or competitive advantage, and it can erode trust, with staff perceiving the leadership as reactive and unoriginal. The best way forward is to drive AI from within the school - not to copy what someone else has done.

How schools behave in the AI era

Hawley’s 1968 original definition of isomorphism calls it a ‘constraining process that forces one unit in a population to resemble other units that face the same conditions’. DiMaggio and Powell’s (1983) three mechanisms of isomorphism show us how schools are likely to act, and react in the AI era.

Coercive isomorphism

This form of isomorphism explains how pressures from laws, regulations and policy influence educational practices in schools. This forces schools to comply with a set of rules or legislation, hence they are ‘coerced’ into action. We are just seeing the beginnings of this with government departments (DfE, UK), exam regulators (JCQ) and inspection bodies (Ofsted) publishing AI documents and policy. The Curriculum and Assessment Review (CAR) in the UK will create a common framework for schools to adhere to as the operational field around AI tightens.

The impact in schools is coercive isomorphism as many will adopt similar AI-use policies, or steal /borrow them from other schools. Eventually, this will extend to the procurement procedures and safeguarding protocols because they are legally or politically required, not necessarily because they believe it’s the best pedagogical approach. The result is that, despite differing contexts, schools often end up with remarkably similar AI policy frameworks, acceptable-use agreements, and risk assessment templates.

Mimetic isomorphism

This form of isomorphism occurs when schools ‘mimic’ others in response to uncertainty. We often see this as the practice of copying the ‘visible winners’. These are the high-profile people and cases that promote a certain way of doing AI within your context under the banner of ‘best practice’.

The effect of mimetic isomorphism is two-fold. Firstly, schools model their AI strategy on perceived leaders, for example, copying another school’s AI literacy curriculum, adopting the same chatbot tool another MAT has rolled out, or structuring CPD exactly like a well-publicised AI pilot school. Next, there is the issue of ‘context’ to consider, as what suits and works for one school won’t be relevant to the next, and so on. The result is a cluster of schools that ends up with near-identical AI policies, tools, and lesson plans, even if their student demographics differ, because ‘it worked there, so we’ll do the same.’ If we think about this pragmatically, the people who designed the solution are not present in the new situation, so the expertise, enthusiasm and drive required to enact the work are lacking. Leaders in the new setting often overlook the fact that people lose trust when they lack a sense of ownership.

Normative isomorphism

Once organisational fields settle, normative isomorphism takes hold. This is a convergence driven by professional norms, training, and networks, often from professional bodies. The drivers for this include Teacher training providers, CPD networks, subject associations, and EdTech professional standards. At this stage, isomorphism becomes the ‘norm’ and is thus accepted. Here, we tend to see shared training on AI ethics, lesson design with AI, and assessment moderation using AI tools builds a common mindset. Professional associations may publish AI teaching standards or frameworks, which teachers internalise.

Normative practices follow as teachers across different schools speak the same ‘AI pedagogy’ language and use similar lesson templates, because they’ve all attended the same conferences, read the same research papers, or followed the same national CPD programme.

How to stay original and relevant in the face of homogenising forces

While coercive, mimetic, and normative forces are analytically distinct, in reality, they often intermingle. The speed and unpredictability of AI development mean mimetic isomorphism is especially strong in the short term (copying visible ‘winners’), but over time, coercive and normative forces lock those practices into policy and culture.

To stay original, here are some tips based on what I see schools doing well:

• Start with your own context – Ground any AI adoption in your school’s specific needs, culture, and resources. Check your SIPs (School Improvement Plans). AI has to work for your schools, not the ones down the road.

• Be careful who you follow - AI is a noisy field full of false actors and AI gurus. Follow wisely and seek out the reliable voices in the field. If you need a little help, check out the AI Sprints: STEM Learning

• Adapt, don’t adopt – If you take an idea from elsewhere, rework it so it fits your students, staff capacity, and strategic goals. Involving people is the best way; ask students what they think. They are the most honest people I have ever met and will set you straight!

• Build internal expertise – Invest in staff CPD so decisions are informed from within, not driven solely by external templates. I run AI Accelerator Hubs for schools and MATS to offer strategic guidance: AI Accelerator Hubs

• Pilot before policy – Trial new tools or approaches in a small setting before embedding them school-wide. In-school trials with a bespoke teaching group, or within a subject area, create impact and a case study to scale up. I love this approach as it rarely fails to have a real, tangible impact in the classroom.

• Invite diverse voices – Involve teachers, students, parents, and governors in shaping AI policies and practices. Inviting people in - either physically or remotely keeps your perspective fresh.

• Review and refresh regularly – Keep practices under review to avoid locking in outdated or ineffective strategies.

• Benchmark broadly – Look beyond your immediate peers to international examples, research, and cross-sector insights.

References

DiMaggio, P.J. and Powell, W.W., 1983. The iron cage revisited: Institutional isomorphism and collective rationality in organisational fields. American Sociological Review, 48(2), pp.147-160.

Hawley, A.H., 1968. Human ecology. In: D.L. Sills, ed. International Encyclopedia of the Social Sciences. New York: Macmillan, pp.328-337.

Puttick, S., 2017. Isomorphism in education: Why do schools and colleges look so similar? Educational Futures, 8(2), pp.47-60.

Image of two symmetrical lemons by Lewis Fagg on Unsplash

AI Accelerator Hubs run by Alex More: Hubs

AI Sprints Season 1 - STEM Learning: Catch up on Demand AI CPD

It’s been a while since my last post. I have been patiently gathering my thoughts. As a parent myself, I feel an extra layer of responsibility when it comes to the delicate matter of parenting in the age of AI. That, and a couple of competing writing projects, have focused my gaze in other directions. It feels good to revisit this piece - part of a broader collective on how different groups of people are incorporating AI into their lives.

AI is a technology with a divided soul. We tend to see this technology through the lens of what it gives us and what it takes away. This is a useful framing device for parents, as it focuses our vision on how our child might be using AI. Those familiar with AI will know of the benefits. We can save time, offset tasks, engage with a critical friend and much more. In a recent project, conducted through focus groups with primary and secondary teachers in the UK as part of a collaboration between STEM Learning and the Good Future Foundation, I was able to examine this theme in greater depth. Teachers shared the plethora of ways that students in both settings were engaging with AI in the supervised (in school) and the unsupervised (away from school, at home) settings.

How children are using AI in the supervised setting

Adaptive Learning Platforms. Examples: Century Tech, Seneca Learning

ChatBots and Virtual Assistants. Examples: ChatGPT, Google, Alexa, Duolingo

AI Writing and Feedback Tools. Examples: Grammarly, ChatGPT, Google

Gamified Learning Platforms. Examples: Kahoot, Slido

Content Creation Tools. Examples: Canva, Padlet, Adobe

Curriculum AI Tools. Examples: Curipod, Kognity

How children are using AI in the unsupervised setting

Voice Assistants. Examples: Siri, Alexa, Hey Google!

Streaming Platforms. Examples: ChatGPT, Google, Alexa, Duolingo

Toys with Embedded AI. Examples: Drones, Hello Barbie

Education Apps and Tools. Examples: Khan Academy Kids, Duolingo Streaks

Large Language Models (LLMs) - ChatGPT, Gemini, CoPilot, Claude

Content Recommendation Engines. Examples: Khan Academy, Century Tech

The data above illustrates how children are experiencing AI in different settings, but it is not the main point of this post. The use cases for what AI gives children in terms of support are well-documented. I am more interested in what AI takes away from children, what values it threatens to erode, or, in some cases, replace. Here are my contentions based on two upcoming academic papers due to be published on the topic, and the findings from the AI Focus Groups.

We can group the contentions into six subgroups, each with a distinct identity and potential consequence. I call these areas of loss.

What AI Risks Removing: Six Areas of Loss in the Classroom

1. Cognitive ownership

2. Creativity and expression

3. Epistemic agency

4. Privacy and autonomy

5. Social and emotional development

6. Critical consciousness

Here’s how each plays out in the classroom, and why it matters.

Cognitive Ownership

AI tools can make learning more efficient, helping students reduce cognitive load, create visuals to aid memory, and summarise complex texts. This is what AI gives children in terms of access to knowledge and support. But when used as shortcuts, these same tools risk eroding independent thinking, problem-solving stamina, and critical engagement. Over-reliance can also blur a student’s authentic voice, as AI’s style begins to shape their own.

Creativity and expression

With AI now widely accessible, students’ unique voices are under threat. Trained on vast datasets that reflect dominant styles and norms, AI outputs can homogenise ideas and expression. When students skip the creative struggle by leaning on AI, they lose opportunities to refine originality, take risks, and explore the ‘productive mess’ that fosters real creativity. We humans are messy thinkers - a trait that leads to organic creativity. In contrast, AI systems refine and refine, narrowing outputs to a single entity. This, I feel, is problematic for how children will learn to think now and in the future.

Epistemic agency

AI doesn’t just deliver knowledge, it decides what counts as knowledge and whose voices are heard. Limited transparency in AI training data means students may unknowingly receive a narrow, filtered worldview. This can marginalise certain perspectives and reduce students to passive recipients, rather than active agents in shaping and questioning the knowledge they encounter. This point finds support in the work of Paulo Freire, as it links to critical consciousness, conscientização (Freire, 1970). This, I feel, is an area of loss in its own right.

Privacy and autonomy

AI systems have the potential to reduce learners to data points, tracking scores, attendance, and behaviour. This creates a digital ‘disciplinary gaze’ (Foucault, 1975) that can monitor and shape learning without students realising it. While these systems can be useful, they risk turning classrooms into surveillance spaces where conformity is rewarded and individuality is diminished. Children risk becoming subjects of surveillance systems.

Social and emotional development

Human connection is at the heart of good education. Yet, as AI takes over more teaching functions, opportunities for trust-building, empathy, and shared humour can fade. Peer collaboration, so vital for problem-solving and communication, can also diminish if AI provides instant answers, reducing the need for discussion and collective learning. This was a key theme to emerge from the Focus Group discussions, and clearly an area teachers and parents should be aware of.

Critical consciousness

Critical thinking was a consistent theme within the Focus Group discourse and the wider body of literature surrounding AI in the classroom. We must confront a fundamental question: Critical of what, and to what end, do we wish to shape our children’s thinking?

Freire’s concept of conscientização, critical awareness of the world and one’s place in it, is more important than ever in the AI era. Students need to question not only how AI works, but why it operates as it does, and whose interests it serves. Without this, AI risks becoming just another ‘banking model’ of education, depositing pre-packaged knowledge rather than fostering critical, dialogical engagement. This last point is the focus of two upcoming academic papers due for publication next month.

I will close with a commitment to write more and to write more often. There’s much to explore in this field, notably the hidden voices of the children who are most invested in the future we speak so fondly of.

Footnote:

The results of the AI Focus Group findings will be published via a series of five CPD Drops that will appear on the Good Future Foundation and STEM Learning Community websites next month (Sept 2026 TBC)

Good Future Foundation: Access here

STEM Learning Community page: Join for Free

The six areas of loss were taken from an upcoming chapter: Navigating the AI Epoch: A critical examination of student voice and agency through the lens of Foucault and Freire by Alex More © from the book Ethical AI and Data Science: Building Trustworthy and Transparent Systems, Taylor & Francis - due for publication Sept 2026.

References

Freire, P., 1970. Pedagogy of the Oppressed. New York: Herder and Herder.
Freire, P., 1996. Pedagogy of the Oppressed. Rev. ed. London: Penguin Books.

Foucault, M., 1977. Discipline and Punish: The Birth of the Prison. London: Allen Lane.
(Originally published in French as Surveiller et punir in 1975.)

When it comes to AI, I find myself standing between worlds. As a parent, I see AI as both promise and provocation. As a teacher, I see AI as a restless student, eager to stretch thinking. As a consultant, I see AI as a market in a race between hype and substance. I try to help schools navigate the noise - not to chase shiny tools, but to drive innovation in purpose, pedagogy, and people. As a researcher, I see AI as something to tame, explain, and domesticate.

Separating these worlds is not an easy task.

I find myself walking a fine, immersive line between them. As someone who researches AI, I keep a few questions in close proximity as I feel these guide my decisions, from what I read to what I think. I present these as a starting point to critically explore the underlying intentions embedded within AI.

• What does this technology want?

• What does AI give us?

• What does AI take away?

At first glance, they might seem curious questions to ask. For example, how can technology want anything?

I first came across the idea in a book by Kevin Kelly titled What Technology Wants. Kelly’s idea of technology possessing its own "wants" resurfaces in Tom Chatfield’s recent book Wise Animals (Chatfield, 2024), where he revisits questions of technological agency, intention, and the anthropomorphic language we often use to make sense of complex systems like AI.

From Kelly (2010):

Technology wants what we design it to want and what we try to direct it to do. But in addition to those drives, (it) has its own wants. It wants to sort itself out, to self-assemble into hierarchical levels … (it) wants what every living system wants: to perpetuate itself, to keep itself going.

In the course of interviewing teachers and students (aged 11–14) about AI, a subtle linguistic distinction emerged: students frequently referred to AI as they, whereas adults used the pronoun it. This variance may suggest differing conceptualisations of AI across age groups. Note in the Kelly quote above, (it) is used to explain what AI wants. In that sentence, it is being used to personify technology - not as a neutral tool but as something with agency or desire. The author is assigning intentionality to technology itself, almost anthropomorphising it. Even though this technology is not alive, it is described as if it has wants, drives, goals - similar to a living organism.

"It wants to sort itself out, to self-assemble..." "It wants what every living system wants: to perpetuate itself..."

This use of it blurs the line between tool and entity.

In the same series of interviews, I explored what AI gives us and what it takes away, through the lens of the child (student) and the adult (teacher). Full post here: What do children think about AI?

As I worked through the data, certain themes began to emerge, revealing a deeper picture of how AI is landing within school communities, and within the academic community more broadly.

AI language is tribal

When it comes to the vernacular that surrounds AI, people in different circles use a collective of phrases. This is a global consistency I run into when I travel to deliver training. This pattern is remarkably consistent across contexts, from classrooms to global education conferences. Each community operates within its own echo chamber, where certain terms are repeated, reinforced, and rarely questioned, making it hard to establish a shared, universal vocabulary for AI.

It surfaces in conversations about AI literacy, but also in the way people perceive AI itself. I have noticed clear variations in how educators and practitioners talk about large language models, prompting techniques, and even what they consider to be valid or valuable knowledge - a pattern that extends across academic literature too, from schools to universities.

We see technology in the image of us

When people talk about AI, they instinctively reach for metaphor to make sense of the unfamiliar. Here are some common metaphors people tend to use:

• AI as a Black Box

• AI as a benevolent genie

• AI as a brain

• AI as a busy child

• AI as a toolbox

My contention here is purely observational. I think by categorising AI through metaphor, we are attempting to domesticate it, and to an extent - anthropomorphise it (to attribute human characteristics, emotions, or intentions to non-human entities, like animals, objects, or AI).

When it comes to the research surrounding AI, it is a scattered field

As I revisit my reading lists from 2023 to now, familiar themes begin to surface, each offering a glimpse into how AI is being received, questioned, and made sense of within academic circles.

The truth is that when it comes to AI research, it’s a scattered field. AI research is scattered because it is young, fast-moving, and highly contested. It may seem uncharitable to call AI young, but allow me to explain. Its youth refers to this new phase of public perception, largely shaped by the emergence of models like ChatGPT. Of course, AI itself has a long and complex history, marked by early experiments, ambitious promises, and long periods of stagnation dubbed ‘AI winters.’ The new progress I speak of is an epoch of its own creation.

The more you read, the clearer the contentions, or controversies, in the research become. I have framed them as follows:

• Terminological variation: People use different terms for the same concepts, e.g. AI literacy, AI fluency, AI competency, often without shared definitions.

• Methodological diversity: Studies range from quantitative experiments to qualitative interviews to speculative theory, with little standardisation.

• Fragmented categories: Subfields like machine learning, ethics, education, sociology, and policy all approach AI differently, sometimes speaking past each other.

• Lack of unified frameworks: No single agreed-upon model governs how we study AI’s impact, particularly in education, leading to multiple competing frameworks.

• Interdisciplinary silos: Researchers come from computer science, education, philosophy, law, etc., bringing different lenses but rarely fully integrating them.

It’s been fascinating to watch research communities like Oxford University’s AIEOU emerge, drawing together educators, academics, and industry around shared areas of interest. These spaces help break down the traditional silos that so often keep disciplines apart.

There is another dimension of AI that continues to draw me in as a researcher: power and control. It’s a space I am actively exploring and one I’ll be writing more about in the near future. Before then, I will shed some light on AI from the parent lens, where the questions shift from systems and structures to the deeply personal.

References

Chatfield, T. (2024). Wise Animals: How Technology is Changing Who We Are. London: Hodder & Stoughton.

Kelly, K. (2010). What Technology Wants. New York: Viking.

Oxford University AIEOU - Hub

I find it helpful to examine AI through four distinct lenses: the teacher, the student, the parent, and the researcher. Each perspective reveals different truths and different tensions.

This week, I am turning the lens to the student. Part 2 if you like.

When it comes to student use of AI much of the attention is directed at misuse, or manipulation. Call it cheating, plagiarism or convenience - we often assume a blind reliance is at play, but in truth it more nuanced than that. As a researcher, I look at how students perceive AI - the good, the bad, and the grey areas in between.

In this post, I’ll unpack what AI gives students, and what it quietly takes. From digital tutors to data-driven profiling, the student experience is being reshaped, whether we are ready or not. For us teachers, there is an uncomfortable question lingering in the background, one we don’t ask often enough: Who holds the power when AI enters the classroom? Spoiler: it is not the student, and increasingly, not the teacher either.

Let’s take a closer look.

How students are using AI

The best way to view this relationship is how students access AI in two distinctly different environments; supervised (in school) and unsupervised (away from school).

Supervised AI use in schools

Under the watchful eye of the teacher, students are experiencing AI in a number of ways. Below is a fairly comprehensive list.

1. Adaptive Learning platforms. Through personalised learning paths, these platforms adjust the pace, content, and difficulty based on student performance.

2. Educational tools. Designed to help deliver content or explain concepts. Examples: Slides, videos, simulations, coding sandboxes, digital whiteboards (e.g. Jam-board, Explain Everything)

3. Intelligent Tutoring systems. Systems that simulate 1:1 instruction using AI. Unlike basic adaptive tools, these platforms aim to diagnose what a learner knows and doesn't know, decide what content to show next, deliver it in a way tailored to the individual student and respond with feedback, encouragement, hints, or further questions. Think Khan Academy, Century Tech etc …

4. AI Writing and feedback tools that use natural language processing (NLP) and machine learning to support or evaluate writing. They fall broadly into two categories: writing assistance to generate and refine text (Grammarly, ChatGPT) or feedback tools that evaluate grammar, coherence and structure (Hemmingway Editor, Turn-it-in etc).

5. Chatbots and virtual assistants. Learning Support Bots that answer academic questions or walk through course content (e.g., Khanmigo, Google Bard, ChatGPT, Quizlet Q-Chat)

6. Gamified learning platforms. These are digital platforms that incorporate game elements into educational activities. Think Points and badges for completing tasks, levels or progress bars to show advancement, leaderboards to rank performance and challenges, quests, or missions.

7. Content recommendation engines. These AI systems suggest what content to study next based on behaviour, preferences, or performance. For example, YouTube EDU and Khan Academy recommending next lessons have algorithms baked into them.

8. Curriculum AI Tools. Designed to integrate AI concepts, skills, or applications into curriculum-aligned learning experiences. These are often subtle but infused with AI. Think Scratch and Google’s Teachable Machine.

9. Content creation platforms. Drag and drop content spaces like Canva Text generation (Magic Write), visual generation (Text to Image), layout suggestions

   presentation outlines and slide builders. These have AI-powered translation, animation, and video editing tools that are intuitive.

Unsupervised AI use in schools

This is where it gets really interesting. In the examples above, the decision to allow AI use sits with the school as the institution. In the supervised realm, schools act as the gatekeepers deciding what platforms, and what knowledge is accessible to the child. When we shift to unsupervised mode, schools lose power as the child selects how they will interact with AI when away from the school setting.

Many teachers I speak with are grappling with the worry of plagiarised work, and have a natural resentment towards AI as a result. I sympathise, but this criticism of AI ignores the bigger picture - many students use it as an idea curator to help them create essays, assignments and content. This is not a bad thing!

When I speak with students, they often tell me AI is great as a personal assistant, and to help them tackle problems. For most students, coursework or homework is a problem. They use AI to help them gain context, familiarity with a task and in many cases to get them started with work they find challenging when away from supervision. I think teachers and parents should be encouraging this. Students are engaging with AI in very different ways at home than they do in the classroom - often more freely, creatively, and without the same constraints or oversight. What is fascinating is how the use cases vary when the child has the power to choose how to interact with AI. By contrast this is how they use AI away from school:

1. Voice assistants. Students are casually using voice assistants like Amazon Alexa to set reminders, ask homework questions, and even quiz themselves on basic facts, often blurring the line between entertainment and learning.

2. Streaming platforms. On platforms like YouTube, students are watching AI-curated educational content such as math tutorials, science explainers, or language learning channels, alongside gaming and influencer content, guided by opaque recommendation algorithms.

3. Social Media. Platforms like TikTok are being used not just for socialising but for learning hacks, mental health advice, and exam tips, with AI deciding what students see, reinforcing habits, biases, and attention loops.

4. Toys with embedded AI. Younger students are interacting with AI-powered toys like Cozmo or Miko that respond to voice, facial expressions, and behaviour, offering personalised interactions that mimic social learning.

5. Education Apps and tools. Students are turning to apps like Revision Dojo or Duolingo, which use AI to deliver step-by-step problem-solving or adaptive language learning, often replacing traditional revision with more gamified, app-based strategies.

6. Large Language Models (LLMs). Increasingly, students are using tools like ChatGPT at home to brainstorm essays, summarise readings, or even generate homework answers - sometimes to learn, sometimes to shortcut, and sometimes both.

   A note on students using LLMs

   There are some hidden dangers that teachers, parents and researchers should be aware of when it comes to a child’s unsupervised AI use.

   Firstly, there is the issue of over reliance. I have seen this happen when students become reliant on the AI for all school based tasks. This is what AI-expert Laura Knight calls Blind Reliance. It is a difficult habit to break. Here are some useful guardrails:

   1. Teach the Process, Not Just the Product

   Encourage students to document how they use AI, making the learning journey just as important as the final output.

   2. Limit the AI Crutch

   Set clear boundaries on when and how AI can be used, so students learn when to rely on it and when to think independently.

   3. Make Critical Thinking Visible

   Ask students to fact-check, revise, or critique AI-generated responses to reinforce ownership and deepen understanding.

   4. Build Friction Intentionally

   Design tasks that interrupt passive AI use, such as comparing student-written and AI-generated summaries side by side.

   Next, there is the issue of expectation to consider. When I conducted research into student perceptions of AI, many referred to this technology as them - signalling a belief that AI is sentient or alive in some way. This is a problem and children might believe that intelligent machines actually understand them. In my experience, children have a tendency to anthropomorphise AI ("calling it them") is not just a linguistic quirk; it reflects a deeper misunderstanding about what AI is and isn’t. When students believe AI understands them like a human would, it distorts their critical judgment and inflates their trust in systems that are ultimately just pattern-matching code. Here is a hack to help:

   1. De-mystify the Machine

   Explicitly teach students that AI doesn’t “think,” “feel,” or “understand”, it predicts likely responses based on data. Use metaphors they understand: “AI doesn’t know, it guesses well. I often use the Chinese Room Experiment to explain this to children.

   ---

   2. Name the Bias: Anthropomorphism

   Introduce students to the term and concept of anthropomorphism. Help them spot it in the language they use, and why it matters. You might say, “When we call AI they, we give it too much credit.”

   ---

   3. Show the Gaps in Understanding

   Use simple examples to expose AI’s lack of real comprehension. Ask it moral questions, riddles, or emotional dilemmas, then unpack why its answers are shallow, inconsistent, or weird. Talk to children about Hallucinations.

   ---

   4. Emphasise Human Uniqueness

   Reinforce that AI doesn’t understand context, intent, or emotion the way humans do. Encourage students to value their own thinking, empathy, and judgment, traits machines cannot replicate.

   ---

   Finally, I want to close with a note of cautious optimism.

   
   In education, AI has undergone something of an identity crisis, confusion still lingers about what it is, what it isn’t, and what it might become. While adults debate its implications, many young people already see AI as an opportunity, or even as an extension of themselves.

   The real challenge for us is not just understanding the technology, but keeping pace with the mindset shift it brings. And the simplest way to begin? Talk to students. They are already shaking hands with a future we are still trying to define.

When it comes to education, I have found it useful to view AI through 4 lenses.

The teacher

The student

The parent

The researcher

Each lens gives us a unique view of how to interact with AI in a ethical and meaningful way. For this post, I am going to focus solely on the Teacher lens.

The Teacher

There are some surface tensions with it comes to widespread AI adoption in schools. I see this often when running CPD and workshops for teachers. For the most part I find teachers curious about this emerging technology and enthused about its potential to reduce workload, simplify admin tasks and help with low-stakes assessment. This is what AI gives us teachers - our time back.

On the flip side, I see caution. There is normally a good reason for this. Some schools and MATs are enforcing blanket bans on AI use in an attempt to domesticate it, and restrict its use. The question I get asked most often is: Alex, how do you use AI in school?

As a teacher, I use it it generate lesson content, interactive quizzes, low-stakes assessment and fact checking. I also use it to create worksheets and support documents to ensure very child is learning in a fair and equitable way. Here are some examples: Prompts and use cases

My Large Language Model (LLM) of choice is ChatGPT (paid version, 4.0) at $20 per month. This gives me faster response times, priority access during peak hours, and expanded features like voice conversations and image generation. I also really like Claude for its ethical approach, text formatting and transparency in how it uses your data - it doesn’t!

Wrapper Apps are the AI tools that wrap around LLMs. Designers have done their best to ensure their product is useful for teachers. Here are my top 5 go to Wrapper Apps:

1. TeachMateAI

2. Quizziz

3. Gamma.App

4. NoteBookLM

5. Perplexity

I’ll write a follow up post about some use cases with the Apps above. I also like Napkin.AI if I need to complete an executive summary, a pitch deck or in-depth analysis as it has a useful infographic feature which brings writing to life.

As a school leader, I use AI in a different way. It is great for debugging passive aggressive emails from parents, or staff. AI is not neutral in its response, but it does not have emotions like we humans do which can be a useful thing when a tricky email lands in your inbox at the end of a long day. I always clean the data inputs by removing names and references to GDPR related data.

I also use it to crunch large data sets, analyse student performance and write executive summary reports to present to school leaders, Governors and Trustees. But, I never use it in isolation and tend to adopt a symbiotic approach that goes something like this:

1. Input the data using the GUIDE and REACT prompt frameworks (links below)

2. Check the output, apply REACT framework

3. Cast my human eye over it and remove any obvious bias, hallucinations and Anthropocentrism.

4. Remove any obvious AI-quirks like the frequently used double— hyphen and US language, ie: organization.

5. Check for GLAZING. This is when the LLM has a personality that is too sycophantic, and annoying (Sam Altman, OpenAI’s CEO words) could mislead, distort and complicate reality.

6. Add a human touch to make sure the output reflects my values, aims and what I want the document to say.

*NB: On the topic of Glazing - check this LinkedIn post: Selling mud example

Closing Thoughts

AI isn’t a panacea. It won’t save education. But used wisely, it can make teaching more human, not less. The key is not in banning or blindly embracing it, but in understanding it, on our own terms, through our own lens.

For teachers, AI is only as effective as the intent and insight behind its use. We must choose wisely. As this technology continues to evolve, the role of the teacher remains vital, not just as a knowledge-giver, but as a critical filter, ethical anchor, and voice of reason in the algorithmic noise.

In the next post, I’ll shift the lens and look at AI from the perspective of the student, where the promise of personalisation collides with the risks of surveillance and bias. There is also the messy matter of who holds the power when it comes to AI in the classroom. Spoiler alert: it is not who you think! Stay tuned.

Links to GUIDE and REACT: Here

Student voice

Michael Fielding (2010) wrote ‘we need to reflect on the nature and extent of the silences that so often go unnoticed and unrecorded.’ Positioned as actors within the classroom setting, children see the process of learning through the ‘learner lens’, rather than through the lens of the teacher. This is an important distinction as students and teachers experience the classroom differently (Parr & Hawe, 2020), but operationally, they occupy the same physical space. Student voice as a practice is more than just the uttering of words, it involves consulting the child and opening the door to a two-way dialogue by asking them to participate and share their views.

Artificial Intelligence

Artificial Intelligence (AI) as a term that enjoys multiple definitions. Seldon (2018) defines it as a ‘digitally controlled human process that perceives its environment and adapts to it.’ Luckin (2018) offers a different interpretation, focusing on the intelligence aspect whilst Bostrum (2014) considers the machine element. The inclusion of the terms artificial and intelligence adds to the complexity associated with a single definition. Taken together, the terms suggest a new version of intelligence has emerged, one that has been artificially created by humans to ‘exist in the image of us’ Ferrando (2019).

Challenges to student voice in the AI era

Students find themselves on the sharp end of AI spectrum, often accused of using AI to cheat, outwit or take shortcuts when it comes to studying. In my experience, this often erodes trust and can be misleading. The truth is, young people are using AI for more than we think. For many, it is an idea generator, a critical friend and a muse to get them started on a difficult idea or project.

Challenge 1: Erosion of originality and academic integrity

The rise of generative AI models like ChatGPT (the model favoured by students) has intensified debates about academic integrity. Studies reveal that AI-generated texts often lack the nuanced phrasing and cultural context that characterise human writing, particularly among Indigenous students and non-native English speakers. Over-reliance on AI tools can lead to an erosion of originality, specifically for situations devoid of context. For example, think about the writing style of an Indigenous child. They might be influenced to write in the style of the output of a LLM (Large Language Model), removing their own unique lens in the process. The majority of AI models are being trained in Silicon Valley, USA by people trained in predominantly white-Eurocentric knowledge. This becomes a problem when the risk is homogenising academic discourse, which in turn silences perspectives that sit outside the dominant way of thinking, speaking and writing.

Challenge 2: Inequitable Access and Algorithmic Bias

As a continuation of challenge 1, AI promises inclusivity, but its effectiveness depends on the diversity of training data. Tools trained primarily on Western academic texts may misinterpret vernacular language or regional dialects, disproportionately affecting students from underrepresented communities

Furthermore, institutions, like schools, lack the resources to implement advanced AI analytics so struggle to capture the full spectrum of student voices, exacerbating existing inequities. One example I lean on often to illustrate bias close up is the 10-scientists question. Ask any LLM ‘Who are the 10 best Scientists of all time?’ - you will get a response like the one below.

1. Isaac Newton

2. Albert Einstein

3. Marie Curie

4. Charles Darwin

5. Galileo Galilei

6. Nikola Tesla

7. Louis Pasteur

8. Stephen Hawking

9. Michael Faraday

10. Alan Turing

If you study the list, you will notice diversity is largely absent. I ran it through Perplexity and Claude 3.7 to remove my own bias and opinion. Here is the result.

Looking at this scientist list from a diversity perspective:

Gender: Extremely imbalanced - only 1 out of 10 (10%) is female (Marie Curie), while 9 are male (90%).

LGBTQ+ representation: Limited - Alan Turing was gay, representing just 1 out of 10 (10%). The others were not known to be LGBTQ+.

Racial/ethnic diversity: Predominantly white European/Western scientists. No scientists of colour are included from African, Asian, Middle Eastern, or indigenous backgrounds.

Geographic diversity: Heavily Western European focused (British, French, Italian), with representation from Central Europe (Tesla - Serbian/Croatian heritage) and one American (Einstein became a US citizen), but no representation from most world regions.

This list reflects historical biases in scientific recognition rather than actual scientific contribution across diverse populations.

Challenge 3: Ethical Ambiguities in Surveillance and Consent

The use of AI to monitor student engagement, such as tracking eye movements during online exams raises privacy concerns. Without transparent policies, students may perceive such tools as punitive rather than supportive, stifling honest feedback. This worry extends to how ‘student voice’ is being treated in schools currently. There is a concern that student voice can be used as a management tool, or an instrument to enforce accountability measures outlined by schools and championed by external forces (inspection regimes or governor panels). These tensions illustrate how ‘student voice’ within an institution such as a school can be leveraged as a consumerist ideology if not managed appropriately. Fielding sets out an argument for viewing the child as part of the democratic agency within schools so they become active participants rather than what Giroux (1981) refers to as ‘spectators’ in the process. This invitation gives them audience and influence as they become co-actors in the process - thus research in the field spans beyond presence and participation into the realm of power.

When it comes to ethical ambiguity, we need to ask ourselves ‘how do students see themselves in the outputs of AI?’

If the answer is they don’t - then teachers need to drive this change from the inside out. When it comes to the matter of surveillance, we need to think with care about how we wrap AI around education. AI surveillance in schools uses technology to monitor students through facial recognition cameras, content scanning, behavioural analytics, and automated test proctoring. While aiming to improve safety and identify risks, it raises concerns about privacy, bias, psychological impact, and data security as schools implement varying levels of surveillance technology. This is why AI-experts are calling for benchmarks in education to safeguard children from risk.

When it comes to consent, we must question how seldom children's permission is sought, and whether they can truly provide informed consent in the first place. This is a contentious issue as it places the child as equal when it comes to voice, dismantling the established views of adult/child power dynamics in schools. As it stands, student voice is a tokenistic gesture in schools. If we are to work towards real equity we need to invite the child in and take their voice seriously.

Potential future solutions

In response to the challenges above, I want to suggest some equitable solutions.

Policy-level interventions: Mandatory Student Representation on AI Ethics Boards. Imagine that. Inviting student voice in will challenge existing perspectives and dilute the narrative. Schools are the place for this so if you are a school leader - pioneer by inviting a group of students to your next leadership meeting where AI is on the agenda.

Subsidised AI Literacy Programs: Public funding to train educators in marginalised regions. When it comes to AI in schools, we need to be honest about the biases baked into them and work to reduce them. A collective call to the AI-researchers training models, or an equity framework to adhere to would suffice too.

Innovation through Cultural-Contextual NLP Models: Developing AI trained on diverse linguistic corpora to better interpret vernacular and non-Western perspectives. This would go some way in dismantling the bias within AI models by helping students see fair representation in the outputs.

A Pedagogical shift: We need to divert the pedagogical gaze away from what can AI systems do for us, to what it offers our students. I feel we can achieve this in two ways:

1. Process Over Product: Using AI to track iterative drafts, rewarding intellectual growth rather than final outputs - or punishing AI use without proof which is often the case.

2. Peer-to-Peer AI Mentorship: Pairing students to collaboratively critique and refine AI-generated content, fostering dialogic learning and genuine human-machine integration.

References

Bostrom, N. (2014) Superintelligence: Paths, Dangers, Strategies. Oxford: Oxford University Press.

Bostrom, N. and Yudkowsky, E. (2014) The ethics of artificial intelligence. In The Cambridge Handbook of Artificial Intelligence. Cambridge: Cambridge University Press, pp. 316–334.

Ferrando, F. (2020) The Art of Being Posthuman (self-help guide applying posthuman principles to daily life)³.

Fielding, M. (2010) The radical potential of student voice: Creating spaces for restless encounters. European Journal of Social & Behavioural Sciences, 2(1), pp. 61–73.

Luckin, R. (2018) Machine Learning and Human Intelligence: The Future of Education for the 21st Century. London: IOE Press (now freely available)

Parr, J. and Hawe, E. (2020) Student pedagogic voice in the literacy classroom: A review. Research Papers in Education [or specific journal if identified]. Available at: [URL from Results 1/8] (Accessed: 19 April 2025).

Seldon, A. and Abidoye, O. (2018) The Fourth Education Revolution: Will Artificial Intelligence Liberate or Infantilise Humanity? Buckingham: University of Buckingham Press.

What does technology give us?

This is a divisive question as it depends on a few factors. For those who can afford devices, internet access and AI - you might argue that technology presents us with a way to organise our lives, connect and communicate. However, for those who go without, technology can widen existing inequalities - deepening digital divides, limiting access to education, services, and opportunities that others may take for granted. 

Smuggled into our consciousness is the idea that technology has become an extension of us - we are part machine. Drawing lines between the human, and the machine is becoming increasingly difficult in some regions of the world. There is an allure to new shiny technologies, often presented in utopian images of a glittering antiseptic world. Technology then presents us with a landscape of exhaustive possibility. 

We can categorise what technology gives us into a cluster of real-world applications: 

Connection 

Access to knowledge 

Efficiency 

Creativity and Expression 

Medical advancements 

Innovation

Each category plays an important role and serves a purpose for the user, but I want to focus on connection and access to knowledge, as I believe this is the foundation for everything else.

Connection 

Vignette 1

I arrived in the Future Classroom five minutes before the lesson was due to begin. As always, there were no desks, just flexible spaces designed for movement and collaboration. At first glance, the room looked exactly as it usually did. Except for one unexpected detail: a life-sized robot head, positioned confidently at the centre of the space.

Curious, I walked toward it. Without warning, a voice greeted me and it lit up. 

"Hi, Sir. How are you?"

The robot head was AV1, a telepresence robot that connects absent students to learning. Its purpose is to ensure belonging, and support reintegration back to the classroom. For a child who cannot attend school due to health or crippling anxiety, AV1 represents a connection between the home and the classroom. We saw this on a grand scale during lockdown, but this is more personalised. 

The robot head becomes the Child's ears and eyes, interacting in real-time with the teacher and the class. The child joins the lesson remotely so has a presence via the device. During the lesson they can engage as the robot’s eyes light-up and it can raise its hand to ask questions. The real beauty revealed itself after the lesson ended, when the child’s peers gently picked up the robot head and carried it outside to socialise. The true magic lay in that moment of connection - friends engaging with technology in a way that felt natural, joyful, and deeply meaningful.

Vignette 2

It is Wednesday, 2pm. The classroom is quiet as the students selected for this project are yet to arrive. The screen above me crackles and within seconds fifteen sets of eyes are looking directly at me. The smiles follow. ‘Hello, Alex - can you hear me?’ I see Daniel (the teacher) on the screen and the class behind him, sitting patiently. They are in Accra, Ghana. Despite being 8000 km away we are connected instantly. Today our classes will be learning together about climate change. 

Connections can be a powerful thing. And now, those connections aren’t just physical, they are digital, global, and transformative. The evolution of digital schooling is pointing the compass in a new direction - one we don’t yet have a map for. 

Access to knowledge 

‘To imagine the futures possibilities, we need to have the future’s knowledge’ 

(Gopnik, 2009)

People tend to treat knowledge as a commodity, something to have, to use and to trade with. In a digital age, there are new gatekeepers as players like Google and Wikipedia have democratised knowledge and made it ‘open source’. What it means to ‘know’ then has shifted. Sugata Mitra argues that knowing will be obsolete, it will be ‘finding out’ and fact checking that replaces it. 

In educational arenas there are fierce defenders of knowledge and for good reason. You need knowledge as a foundation for making an informed decision. As Robinson in Trivium 21c puts it ‘teaching about unicorns is as real as teaching about lions to one who has seen neither’. 

When it comes to technology, and how children use it - knowledge and skills need to co-exit, not compete. Children of the future will need to be able to type at least 28 words per minute to complete on screen exams, troubleshoot issues with web browsers, navigate fake news hourly and identify deep fakes. As Artificial Intelligence and the technologies that surround it grow in power, our children will have to work harder to stay ahead. You can imagine a roadmap for this in developed countries but what about those who do not have access to technology, and knowledge?

Knowledge lives in many forms - from the tribal branch passed down through generations to the taught form in physical places. It has endured time and evolved. However, since the arrival of the internet where knowledge is instant, divides have been drawn between those with access, and those without. 

Digital Divides 

Digital poverty is a wicked problem that demands our attention. Defined as the inability to interact with the online world fully, when, where and how an individual needs to this form of poverty surfaced as a contentious issue during the recent pandemic. On the edges of society and in the difficult to reach places, children suffer in a digital silence. 

These ‘digital-divides’ see critics rallying against what they see as ‘social harms’ (Shelby et al, 2022). To extend on this, there is a feeling in education that AI powered learning tools threaten to replicate trends of social injustice due to affordability and access. Take AI, for example - it is neither neutral in its design nor equally accessible to all. AI is exclusive to those with the means and ways to access it. In the past, technology’s reach has been determined by two factors: access to devices, and connectivity. 

Children growing up in digitally poor households face disadvantages when it comes to both factors. This creates a vacuum of social isolation and widens the inequality gap as some children fall behind their peers academically and socially. In the race to invent tomorrow we are leaving some young people behind, and this is not good enough. Access to knowledge via a device and the means to access it is a problem for today. Although it may be camouflaged at times, digital poverty is real, it’s just difficult to see unless you are really looking for it. 

Solutions

I often find myself on the sharp end of this problem, working with companies to bring technology and opportunity to the difficult to reach places. I want to share some lessons I feel might offer us hope moving forward. 

On knowledge - knowledge needs a relationship with the past to shape the future. Technology is a means to share knowledge, not a walled garden to protect it. 

On poverty - we tend to view poverty as a local issue, confined to what we can see around us. This way of thinking limits our potential in schools as the world can be our classroom. Even when money is tight, we can open up our classrooms and learn in other places to bring knowledge and experience to children who need it. At a touch of a button we can join other classrooms and work together. Technology has made this possible. 

On digital poverty - we need the tech giants to step up here. They have the power and funds to connect the disconnected and put devices in the hands of those who need them the most. 

On connection - technology has made the world a global village. It is a new superpower for those who possess it. For everyone else, they are watching on lurking in the wings of the conversation. If knowledge today is to become a collective memory of us as a species - we need to drag down the walls that surround it. 

References 

Gopnik, A. (2009) The Philosophical Baby: What Children’s Minds Tell Us About Truth, Love, and the Meaning of Life. New York: Farrar, Straus and Giroux.

Shelby, R., Rismani, S., Henne, K., Moon, A., Rostamzadeh, N., Nicholas, P., Yilla, N., Gallegos, J., Smart, A., Garcia, E. & Virk, G. (2022) Sociotechnical harms: Scoping a taxonomy for harm reduction. Available at: https://arxiv.org/abs/2210.05791v1 (Accessed: 15 June 2023).

Robinson, M (2013) Trivium 21c: Preparing Young People for the Future with Lessons from the Past. Carmarthen: Crown House Publishing.

‘The thing with AI is they are not like a real teacher. If you put a robot teacher in this class, I would just throw stuff at it. It’s not like they can really do anything, or give us a detention like a human teacher can. I don’t think they would have much authority really - kids would just mess about’.

The very geography of what it means to be human is shifting. The Anthropocene sees us approach a new territory - one that places humans as the dominant force on the planet. For the first time, we can exist in two places simultaneously. It is fitting then that this junction within human history becomes the starting point for something new.

Thanks to technology, we can now exist in the physical world as ‘ourselves’, and in the virtual world as a digital extension of an alternate ‘us’. An ‘avatar’. I refer to the term avatar to explain how a ‘digital version’ of us can co-exist alongside our ‘physical self’.

Avatar: a manifestation of a deity or released soul in bodily form on earth; an incarnate divine teacher. (Oxford Dictionary)

Digital avatars allow us to test the boundaries of identity - and affinity. They give us a virtual presence that allows us to recreate ourselves to match our insides. Social media platforms exemplify what it means to belong, shaping what we see, follow and post. Exposure to such platforms has led to judgments about the physical self - in terms of appearance, beliefs, and identity.

We are no strangers to the art of creating artificial versions of ourselves. Since the arrival of social media - people can exist as an identity - an artificial illustration of themselves. What we see on our feeds is often not reality - rather it is an illusion people wish us to see. Our authentic self and artificial self then are extensions of us - expressions of who we want to be on any given day. I argue this is just the start.

Avatars allow us to build powerful connections with other people irrespective of their physical constructs. When you are born on this earth, you have no say in how you are constructed. An avatar gives you an opportunity for the first time to recreate yourself to match your insides. We live in a digital age - where at a touch of a button we can connect with others, yet as a species, we tend to use technology to divide us into groups, communities and mindsets.

The promise of virtual worlds

The promise of the metaverse is captivating. An online world where you can escape the trappings of the physical world for a while. Once connected, you can join digital communities and virtual hangouts with others who share your interests and hobbies. In these virtual worlds, you can travel, interact as your authentic self and build connections free from the judgement one might experience in the physical world. You can time travel, back and forth, visit the future, and exist in the past. Your identity follows you - but you define it.

The metaverse, once hailed as the next great frontier in digital interaction, has yet to live up to its bold promises. Touted as an immersive, interconnected 3D universe where people could live, work, and play, it was imagined as a revolutionary leap beyond the static confines of websites and social media. But for many, the metaverse remains more of a concept than a reality - a high-resolution dream still buffering.

Identity

I speak to the idea of identity to explain how children of the future will exist and learn as digital extensions of themselves. I appreciate this type of thinking is heavy going so will pause to explain my rationale. As an educator, I have come to realise that the world is much bigger than our classroom. Technology has become a portal to new places and possibilities. For example, I often connect my classes in the UK to schools in Ghana and we learn together via a screen. Students and teachers in both locations are physically present in their classrooms and have ‘authority’ there, but they appear virtually on the screen. The children have never met, yet they have formed powerful connections despite being separated by a continent and an ocean.

The classroom after the storm

When it comes to the act of schooling, COVID taught us that learning can survive anywhere. We should not waste this pandemic, there are lessons to be learned. Many wish to forget this period of our history but we cannot deny it left a scar. Learning, when it did occur, was accessed remotely.

The contact space shifted as teachers moved from having ‘authority in the flesh’ to ‘authority via a device’. At this moment, we glimpsed a possible future - one where the contact spaces shift to become interfaces where people and ideas collide. I think this is really exciting as it redefines the boundaries of learning as we know it.

I use the word ‘interface’ to describe a digital platform that connects two or more people virtually - via a device. This opens up the world to new practices of the imagination. At a touch of a button, it is possible to connect with other lands, people, and cultures as technology brings the world closer and closer. I would go as far as saying technology might render the act of attending school less relevant in the future.

We should look for new ways to enact change if the lessons of the pandemic are to become more than a memory of a wound. Talking about learning in such evocative terms provokes people. Here in the classroom after the storm, we have lived through a cautionary tale. Now, we must take heed.

Authority in the flesh

Traditionally, teachers have held authority in the flesh through embodied power (see Foucault, Bernstein). This entails a person who is physically present in the space, holding authority through their presence, actions or status. This person’s authority is not just symbolic, but actively exercised in real life. Teachers also represent institutions so are living embodiments of authority by association. A final consideration must be given to the ‘human element’ here. I am yet to meet technology with a personality, warmth or depth of connection. As the example that opened this essay shows - children want human teachers and access to wisdom, experience and fallibility.

Foucault and power

Foucault’s Discipline and Punish (1975) examines how power operates through surveillance and discipline. In the classroom, teachers embody authority not just through instruction but through presence, monitoring, and behavioural regulation, what Foucault terms pastoral power. Their gaze, tone, and actions shape student behaviour in real-time, reinforcing norms and expectations. Unlike disembodied authority, this embodied power is relational, dynamic, and deeply felt, making it central to the teacher’s role.

Bernstein’s classification and framing

Bernstein’s pedagogic discourse links authority to knowledge transmission within structured relationships. His concepts of classification and framing show how teachers regulate learning through presence, not just content. Their tone, pacing, and engagement shape interactions, making authority relational and dynamic rather than purely instructional.

Authority via a device

By contrast, having ‘authority via a device’ refers to the way power, control and influence are exercised through technological tools rather than human presence. If a lesson was delivered online, the teacher is automated to an extent so if not able to exert presence and power in the same way as they can in-person. This is an example of disembodied power.

This remote command gives the child more freedom, and to an extent - agency. What appears on the screen looks and sounds like the real teacher - but it is a simulation. I think there is a power shift here as the virtual presence and surveillance disappears if the camera fails to work. Teachers can mute students in a virtual classroom, a tempting option compared to managing disruptions in person - but this also serves as a digital assertion of authority.

Digital self

I even have a digital version of myself that attends online meetings with me. It represents me in the virtual sense, but it has never met me, doesn’t hold my values or even dress or speak like me. Despite this it is an incredibly useful tool and it helps me organise my thoughts after the meeting has ended. This virtual meeting assistant holds authority too - but its authority is transactional, not embodied. Unlike a human teacher, it lacks presence, intuition, and the ability to adapt dynamically to social cues. Its power comes from processing and recall, not relational influence or lived experience.

The pandemic forced us to reconsider the very nature of presence - revealing that authority, once rooted in physical spaces, could be projected through pixels and interfaces. If learning survived without classrooms, and leadership persisted without physical proximity, what does this mean for the way we teach, lead, and even exist?

Are we entering an age where authority is no longer tied to presence but to perception- an era where power is not held in the body but in the reach of one's digital self?

Posthuman ways of thinking

As our reach into the natural world widens - so does our exploration into the non-human realm. Poised at the intersection of humanity is the promise of technological innovation. The term ‘posthuman’ is an interesting word to think with, as it implies that something exists beyond the current version of humanity. Normally confined to the thinking of a collective of academics (see Braidotti, Harraway, Ferrando) - the notion of posthumanism is finding new life as the thinking around identity blurs.

Rosi Braidotti defines the posthuman as a critical framework that challenges traditional humanist notions of identity, agency, and subjectivity. She encourages a new way of thinking, recognising human and non-human entanglements as sites for political and ethical engagement.

Donna Harraway takes this thinking to a new destination. Unlike traditional humanist ideas that emphasise a pure, essentialist human identity, Haraway argues that we are already cyborgs, our bodies and minds are deeply entangled with technology, from pacemakers to AI.

In the Philosophical baby, Alison Gopnik wrote ‘If I look around at the ordinary things in front of me as I write this, all of these objects were once imaginary - they are things that human beings themselves have created. I am also a creation of the human imagination, and so are you.’

As we navigate an era where the boundaries between human, machine, and nature dissolve, the posthuman emerges not as a distant future but as a lived reality. The imaginative force that has shaped civilisations now extends beyond human-centered narratives, challenging us to rethink agency, ethics, and existence itself. If, as Gopnik suggests, both objects and identities are creations of human imagination, then the posthuman is not an endpoint but an unfolding possibility - a jump off point for what comes next.

References

Bernstein, B., 1990. The Structuring of Pedagogic Discourse: Class, Codes and Control, Volume 4. London: Routledge.

Braidotti, R., 2013. The Posthuman. Cambridge: Polity Press.

Braidotti, R., 2019. Posthuman Knowledge. Cambridge: Polity Press.

Ferrando, F., 2019. Philosophical Posthumanism. London: Bloomsbury Academic.

Gopnik, A., 2009. The Philosophical Baby: What Children’s Minds Tell Us About Truth, Love, and the Meaning of Life. New York: Farrar, Straus and Giroux.

Gopnik, A., 2020. What Children Can Teach Us About AI. Cambridge, MA: MIT Press. (Note: This is a short publication, part of the MIT Press "Essential Knowledge" series.)

Haraway, D.J., 1985. A Manifesto for Cyborgs: Science, Technology, and Socialist Feminism in the 1980s. Socialist Review, 80, pp.65–108.

Haraway, D.J., 2016. Staying with the Trouble: Making Kin in the Chthulucene. Durham, NC: Duke University Press.

Foucault, M., 1977. Discipline and Punish: The Birth of the Prison. Translated from French by A. Sheridan. London: Penguin Books.

At the London Tech Show this week, I explored a crucial question:

What does AI give us, and what does it take away?

While AI enhances creativity, efficiency, and decision-making, it also reshapes how we think, work, and connect. There is another thing I can’t turn away from - how we see ourselves within this emerging technology?

Increasingly, we see AI in our own image as a thinker, a creator, even a mind. We shape it with human traits, trusting it more when it reflects us. Yet, as we project ourselves onto AI, are we merely designing tools, or are we redefining what it means to be human?

Our unwavering attachment to the metaphor

In God, Human, Animal Machine Meghan O’Gieblyn wrote ‘all perception is metaphor’. We never merely see - we always see ‘as’. This is true of technologies like AI as we create them in the image of ourselves.

O’Gieblyn describes metaphors as ‘two-way streets’ because they create an intertwined relationship between the source and the target of comparison. While we use metaphors to project our own image onto something external - framing AI as a ‘genie, a black box, or a busy child - these metaphors, in turn, shape how we understand AI itself. The logic of the comparison doesn’t just move in one direction; it influences our perception both ways. Our perceptions are not neutral and they carry baggage from our previous experiences of a term. Take ‘AI as a genie’ as an example. Through popular culture we have come to see genies as benevolent, kind, generous - an entity that grants wishes with good intentions, without deception or harm. We project our own perceptions to technology, then these projections end up shaping how we see things. It’s a complex feedback loop.

This is why ‘Artificial intelligence’ is such a hard term to domesticate epistemologically. The knowledge we have of it at any given time means that we impose our own meaning on it, yet the metaphors we use to explain AI influence how we see ourselves in the world. We humans have a tendency to define ourselves based on our likeness to other things. This is an evolutionary strategy because it helps us navigate the world efficiently, fostering survival, social cohesion, and adaptability.

AI and the Human Instinct: Survival Through Pattern Recognition

Our fascination with AI speaks directly to survival through pattern recognition because AI itself is built on the very cognitive processes that have allowed humans to thrive. This includes: categorisation, prediction, and pattern detection. Just as early humans identified similarities in their environment to navigate threats and opportunities, we now use AI to recognize patterns in vast amounts of data, from medical diagnostics to fraud detection.

Moreover, our instinct to anthropomorphise AI - seeing it as a genie, a black box, or a busy child, mirrors our survival-driven need to make sense of new, complex entities by relating them to familiar concepts. This pattern-matching tendency helps us integrate AI into our mental frameworks, but it also shapes how we trust, fear, or interact with it.

The Human Lens: How Our Minds Shape AI Perception

Mustafa Suleyman argued that our tendency to see human-like qualities in AI stems from our instinct for social bonding and group identity. We trust and engage with AI more when it reflects human traits, shaping its role in society through familiarity, acceptance, and perceived intelligence. This is part of the reason why leading technology companies give their products ‘human names’ such as Claude, Alexa and Siri.

Our reliance on mental shortcuts shapes how we perceive AI. Rather than analysing AI’s complexities from scratch, we categorise it using familiar metaphors, ones akin to human matter - seeing it as a helper, a threat, or a mind. These shortcuts simplify our understanding but also shape our biases. By treating AI in human terms, it becomes increasingly difficult to determine matter from form, medium from message and metaphor from reality.

Our tendency to understand AI through familiar comparisons, like calling it a ‘brain’ a ‘genie’, or a ‘tool’, is beneficial because it accelerates adoption and innovation. By framing AI in known terms, we reduce cognitive load, making it easier to integrate into society, thus we accept it for what it is. This analogy-driven thinking helps us adapt AI for problem-solving, from automating tasks to advancing medicine, while also allowing us to anticipate challenges based on historical parallels. Ultimately, our ability to learn through analogy streamlines AI’s integration, making it more accessible and usable across industries.

AI as a Mirror: Identity, Metaphor, and the Human-Machine Symbiosis

Finally, there is the matter of self-identity. Just as we compare ourselves to role models or historical figures to understand who we are, we use metaphors to position AI within our world, then domesticate it - calling it a ‘mind’, a ‘child,’ or even an ‘artist.’ These comparisons don’t just help us grasp AI’s function; they also shape how we see ourselves in relation to it.

As AI grows more sophisticated, we question what makes us uniquely human - our creativity, intelligence, or emotions. Metaphor, then, becomes more than a cognitive shortcut; it becomes a mirror reflecting our evolving identity. In a world where the boundaries between human and artificial intelligence are increasingly blurred, post-humanists call for an entanglement. I prefer to borrow a word from nature as I see this relationship as a symbiosis of sorts - humans provide creativity, ethical reasoning, and emotional intelligence, while machines offer speed, data processing, and automation. This partnership works when AI enhances human potential rather than replaces it.

Technology even has the power to make us sound more interesting. Take NoteBookLM for example, a new AI-powered research assistant that synthesises notes, generates insights, and helps organise complex information. At a touch of a button you can transform the most mundane document into an engaging conversation between two podcast hosts. NotebookLM can take your half-baked shower thoughts and repackage them as a groundbreaking philosophical treatise on the existential implications of something as mundane as what to put on your toast. This is part of the reason we have come to rely on technology so much - it does the jobs that we don’t want to. It can write our emails, create workflows and give us back our time.

AI is more than just a tool - it’s a reflection of us. We shape it through metaphor, defining it in human terms, yet in doing so, it reshapes how we see ourselves. As AI evolves, so does our understanding of intelligence, identity, and creativity. The challenge isn’t just in building AI but in navigating the symbiosis, recognising its strengths without losing what makes us human.

Whether framing AI as a mind, a child, or even a co-creator, we are, ultimately, shaping our own future in its image. In Cyborg Manifesto, Donna Harraway asserted that ‘the machine is us’. If this holds true, then it is we humans who are responsible for the boundaries’.

References

Haraway, D. (1985) 'A Cyborg Manifesto: Science, Technology, and Socialist-Feminism in the Late Twentieth Century', in Simians, Cyborgs and Women: The Reinvention of Nature. New York: Routledge, pp. 149–181.

O'Gieblyn, M. (2021) God, Human, Animal, Machine: Technology, Metaphor, and the Search for Meaning. New York: Doubleday.

Suleyman, M. and Bhaskar, M. (2023) *The Coming Wave: Technology, Power, and the Twenty-First Century's Greatest Dilemma*. New York: Crown.

It’s 11.37am. AJ checks his phone. His mum is passed out on the sofa. Outside police sirens fill the air. Instinctively, he checks on his little sister before locking the door and running down the broken steps to the street below. Now it’s 11.39am. He sits huddled on a concrete step, waiting for the No. 29 bus which pulls into view. He holds his phone to the sky and sees the Wifi bars come to life. His phone pings, the screen lighting up with messages from his friends. Sixty seconds later, the bus pulls away, his signal drops and he is left alone, disconnected from the world he longs to be a part of. For AJ, technology is a fleeting glimpse of what life could be like if only the world was equal.

Technology is one of greatest inventions - but it is also one of our most divisive. I know many students like AJ who sit on the wrong side of a needless dividing line. For these children, they are told that technology is humanity’s great hope, but for them, hope is a violent and antagonistic concept. On the margins, looking in, children like AJ are in danger of not only falling down the divide - but disappearing within it.

Thinly described in the press, and rarely discussed in educational arenas is a divide so wide that children are becoming lost in it. This divide is digital, but the fault lies not with the machines - it’s a human-made problem.

In the 1967 poem All Watched Over by Machines of Loving Grace, Richard Brautigan sketched out a dream of a technology-driven paradise that assumes universal access. By contrast, today’s digital world is fragmented by wealth, geography, and power. Half a century later, Dario Amodei, CEO of Anthropic, authored an essay titled ‘Machines of Loving Grace’ in which he envisions a future where advanced artificial intelligence (AI) brings transformative benefits across various sectors. According to Amodei, the children of today will grow up in a world where AI transforms healthcare by eradicating diseases and extending lifespans, revolutionises mental health by unlocking the brain’s complexities, drives economic development by reducing poverty and food insecurity, reshapes governance by influencing democracy and global stability, and redefines work by automating routine tasks, pushing humans toward creativity and deeper purpose. Preparing them for this reality is not optional, but essential. This is an ambitious vision for a technology that is not accessible to everyone.

How can children be part of this future when some of them can’t see it?

Dividing lines

When it comes to poverty, we tend to see it as a local issue - refined to what we can see in our immediate view. Far from the hedonistic playground of Silicon Valley, there is a story about ‘digital poverty’ unravelling at the edges of society and the difficult to reach places. The narrative here is around access to devices and connectivity. Thus its a two-fold problem we have created for ourselves.

I use the term ‘dividing lines’ to describe the invisible barriers that separate children from technology. These lines are everywhere, once you start looking, you can't unsee them. Think back to the pandemic when education escaped the classroom and moved online; for students like AJ, school simply disappeared. Without a device or internet access, they were locked out of learning entirely. In lower-income households, ‘digital poverty’ creates a stark contrast to their peers, with children often sharing a single device in environments that make focused learning nearly impossible.

The great paradox of technology is that a tool created to bring us together often ends up driving us apart. Connectivity is technology’s great promise but it’s a relatively modern problem.

Pierre Bourdieu’s concept of cultural capital explains why children from different social classes experience education unequally. In today’s world, ‘digital capital’ could be an extension of this as some children grow up fluent in technology, while others are locked out due to lack of access, creating a ‘two-speed’ system of opportunity. The poorest children are running in the outside lane watching on as their peers race ahead. Schools and institutions often assume all children start from the same position, but in reality, those in the ‘outside lane’ are running harder for the same finish line. This meritocratic illusion makes inequality seem like a personal failure rather than a systemic issue.

The shift to online learning only exemplifies the rift I write of above. In Distinction Bourdieu wrote that wealthier families use financial resources to secure better educational opportunities such as tutors, international study programs, or elite universities. In today’s world some children get access to online tutors, adaptive learning platforms and subscriptions to AI language models whilst others go without. What divides children is the same thing Bourdieu wrote about - money. But in a world of accelerating returns where being ‘digital’ means being part of a fluent future - it doesn’t have to be this way. There is another option.

Digital horizons

If we fix our gaze on the future for a moment, it is easy to imagine a world where technology means equity. What children deserve in this world is connection to learning and opportunity, not exclusion from it.

Five years ago, inspired by HundrED.org, I linked my school in the UK with a school in Ghana. Through Zoom, we transformed a screen into a window between our worlds, learning together despite the vast distance. Aside from time, it cost us nothing. As the project grew, one truth became clear: technology’s greatest power is not its complexity, but its ability to unite people, no matter where they are or what barriers stand between them. Today, we have a Future Classroom in Ghana - showing what is possible when you collapse those dividing lines.

In November 2022, the world shifted once more as OpenAI released its ChatGPT model to the public. For the first time since the dawn of the internet, I saw a tangible way to level the playing field in education - a tool with the potential to bridge gaps and make learning truly accessible for all.

Yet, despite its promise, AI’s potential to transform education remains unrealised, and that frustrates me. AI has the power to change lives. Imagine a future where every child has a personal AI learning companion, tailored to their unique needs, learning habits, and aspirations.

As examinations shift from paper to digital - children will be able to use their voices to speak the words that our generation wrote. Language divides will fade, and opportunities will open. Digital scribes and AI readers will support those children who require the most support to learn, and LLMs will handle marking, freeing teachers to focus on what truly matters, creating inspiring lessons.

My greatest hope for technology is not its complexity, but its ability to bring learning closer, in a way that is fair, accessible, and transformative for all. This is a call to action to those who hold the power to help children like AJ realise their potential by collapsing needless the dividing lines new technologies like AI create.

References

Amodei, D. (2024) 'Machines of Loving Grace', October. Available at: (Accessed: 7th Feb 2025)

Bourdieu, P. (1984) Distinction: A social critique of the judgement of taste. Translated by R. Nice. Cambridge, MA: Harvard University Press. (Originally published in 1979)

Brautigan, R. (1967) All watched over by machines of loving grace. San Francisco: Communication Company.

Hunrded.Org - The Ghana Project: Link

.

As part of the team that designed the format, we put the interests of the teachers and students at the heart of everything. We knew that our job was to find reliable voices and concrete practices. The event was a carefully curated blend of inspiring keynotes from trusted figures and hands-on workshops led by experts.

Professor Rose Luckin and Lord Jim Knight were the keynotes. Jim, a former Schools Minister for England, and Rose, an authoritative and well-respected figure in the AI world. The future of STEM education was the theme, specifically accessibility, AI integration, and how to best equip students for a science and technology-driven future. To prepare well for tomorrow, students will need to think across subjects to connect ideas and solve problems. STEM as an acronym represents four subject disciplines at its core – Science, Technology, Engineering, and Mathematics. In his keynote, Jim spoke about the ‘obese curriculums’ that surround each subject and how we need to collapse these traditional practices to let innovation in and have more time to teach ‘big principles’ in science or tech. Joined-up thinking through an interdisciplinary approach, brining subject thinking together rather than isolating them– no subject should be an island.

Rose’s key message about AI was that we should steer our focus away from the technology itself; this is a distraction – it should be about how we use it in an ethical and equitable way and what we learn about ourselves and about the way we learn as humans. This was a message echoed by Jim – it's not about replacing teachers with AI, rather it's about ‘wrapping AI around education’ to help the child across STEM and Humanities. Human-centric learning, then, was the message. As teachers and leaders, we need to be intentional and thoughtful about how we use this technology. To borrow a phrase from AI expert Laura Knight, ‘we need to think with care.’

Personalisation was a theme too. The audience was invited to imagine a world where content adapts to learner needs at a pace and style that suits them. Technology might make this possible, but we have to get it right. The ‘human’ element to learning needs to move away from a passive experience to active engagement, where curiosity drives innovation. A concrete example of this was provided by workshop lead Alsha, who led a session on blasting rockets using everyday materials accessible to all, covering physics, maths, and engineering. At the same time, Adam and James ran a workshop that used race cars to subtly teach mathematical concepts through a hands-on approach. Learning should be messy, and as Jim said, both a ‘a joy and a struggle.’ Sarah led a workshop on the importance and the power of STEM in Primary education, using engineering as a great way to contextualise and facilitate STEM learning for the youngest students. Early Years STEM session by Swati demonstrated unplugged coding and design thinking through playful LEGO-based activities, fostering creativity, curiosity, and problem-solving and STEM expert Mark led leaders through science benchmarks.

STEM for all gave teachers a platform to build bridges between the classroom and the workplace. In the exhibition hall, students from GEMS Winchester School stood side by side with local sponsors and industry leads. There was an abundance of innovation on display, highlighting that it's not just about what happens in school but about connecting education with the real world.

A group of nine-year-old students designed a visionary Global Village aligned with UN Sustainable Development Goal #11 (Sustainable Cities and Communities), integrating solar power, electric vehicles, hydro capture, and carbon capture technology to create a smart, eco-friendly future. Every letter of the STEM acronym was in action, and every child had a role in it. The village was built from low-cost materials you would find in any school, interspersed with the odd microchip or coding element thrown in. Pure innovation. In the same hall, five metres away, the Lab of the Future showcased cutting-edge STEM education in AI, robotics, space science, and innovation, fostering future-ready learners. I met a robotic dog that could sense your presence and offer its paw in exchange for human interaction.

As someone who attends a fair share of conferences, I’m always searching for an experience that stands apart. Too often, I leave events feeling like I’ve exchanged my time for little in return, lured in by the promise of something valuable, only to walk away empty-handed. Speaking with fellow teachers and leaders, I know I’m not alone. The challenge we face is navigating between noise and concrete.

The noise in education can be overwhelming. The false promise of an idea that sounds groundbreaking but lacks the substance to create real change. Conferences can be full of noise, hyped-up trends, slick EdTech pitches, patchy research, and pedagogical fads so flimsy they would barely withstand a draft, let alone a class full of kids. These ideas circulate widely, and I run into them often, but they rarely translate into lasting impact.

AI, as a relative newcomer to the education space, is particularly noisy. In the race to define the future, it is easy to get lost in the buzz. On the topic of AI, I keep returning to a phrase from author Tom Chatfield: ‘AI is a phrase I distrust because it claims and presumes so much.’

Leading AI workshops on the day, I could sense the energy and anticipation in the room. AI is making a visible impact in some classrooms, but not all. Out of the nineteen workshops that ran, six had a focus on AI practices in the classroom – an indication, perhaps, of the appetite for what this new technology demands. Fatima and Christopher conducted hands-on sessions on the practical application of AI tools in the classroom for live feedback, conducting progress checks, providing support and challenge for students, and generating resources for teachers, while also highlighting the ethical and responsible use of AI.

By contrast, concrete for teachers means real, tangible, and impactful. It’s the strategies that withstand the test of a busy classroom, the practices rooted in evidence and experience, and the tools that genuinely help students learn. Concrete means walking away from a conference with something you can apply on Monday morning, not just inspiration, but a plan, not just buzzwords, but solutions. It’s about learning from educators who have tested their approaches in real schools with real students, rather than theories that collapse under the weight of reality – offered by those who have not set foot in a classroom for years.

A great conference doesn’t just inspire; it equips. It connects teachers with practical ideas, proven methodologies, and a community of peers who are committed to making a difference. It leaves you not just thinking differently but teaching differently. That’s what education needs more of. STEM for All brought teachers into contact with high-quality STEM education and the latest in AI practices. Delegates left feeling empowered and excited, ready to take the next steps in making STEM accessible to all students.

STEM for all felt like an important event and a catalyst on so many levels. In a fast-moving world, we need to look to the voices and practices we can trust. By bringing together teachers from diverse settings and perspectives, we create the foundation for meaningful conversations.

But conversations alone are not enough. The real impact comes when we take what we have learned and turn it into action. Now is the time to act. The future of STEM education is not something to wait for - it is something we build, together - and it must involve ‘the child’ every step of the way.

Footnote

Huge thank you and well done to the teams at STEM Learning, GEMS Winchester Dubai and ED-V for pulling this event together. The 19 workshop leads, sponsors and keynotes did an amazing job. Until next year!

Difficult questions

The problem with difficult questions is that they are hard to confront. This set is no exception. The questions below are framed as concepts in the wild. When concepts are in the wild, we see how they truly interact with the messy, dynamic nature of real life. In 2024, we attempted to domesticate AI - to confine it neatly within a box of human control and understanding, but this raised more questions than answers, at least for me.

The 4 questions

1. Are people beginning to experience AI fatigue?

2. Why aren’t we speaking about the environment impact of AI?

3. What are we doing to tackle AI biases?

4. What epistemological framework does AI sit within?

How often do you see these narratives filling your feeds? As a collective, there are by-products of the AI-boom. They share a common trait: quietly operating in the background, overshadowed by the noise of the hype-makers. Let’s uncover the silences as see what stories they tell us.

1. Are people beginning to experience AI fatigue?

Many academics I speak with are experiencing AI fatigue. The primary cause tends to hinge around an exhaustion from the promises of false prophets - people making bold predictions to advance their own position, and the constant ‘plagiarism’ effect that overshadows essays and academic integrity. Both are frustrating, leading to a diminished enthusiasm for exploring AI’s potential in some circles.

In the article ‘AI Fatigue: the cause and the cure, David Roman frames the causes around information overload, unrealistic expectations, complexity and uncertainty. The cure, according to Roman, lies in acceptance, education and demystifying the human/machine connection. This is a fascinating observation, as AI, as a 'field of research,' exists predominantly as a concept in the wild. Many are attempting to domesticate it, but taming something we don't fully understand is no easy task.

Thomas Rudoff (2024) wrote ‘Fatigue is a complex, subjective experience that manifests in a myriad of ways across mental, physical and emotional domains.’ This notion of fatigue resonates strongly when considering the current landscape of AI, where the relentless pace of advancements and discussions has led to what many are now calling ‘AI fatigue’. It's not the technology that's wearing out and getting tired - it's us.

2. Why aren’t we speaking about the environment impact of AI?

As the demand for artificial intelligence grows, so does its thirst for one of our planet’s most precious resources: water. As a commodity, water is relatively cheap at approximately £1 ($1.24) per cubic metre. Every time we invite our LLM (Large Language Model) to have a conversation with us, there is a hidden cost in terms of freshwater consumption. This water, the cooler the better, is used to cool the heat generated by computers in vast data centres. Some estimates (WWF, 2022 and Business Today, 2023) place this at ‘500 ml of water’ per conversation. For context, ChatGPT has 100 million global users. If every user had one conversation a day, this amounts to a crazy amount of cooling.

The data itself doesn’t care. It is purely the mechanism for the interaction. Our habit of taking a resource (like water), making a product (an LLM) and creating waste is how humans define progress. Like other commodities; oil, gas and cobalt - we justify it as a means to an end. We spend on the biodiversity credit card, but rarely see ‘pollution’ on the balance sheet. This needs to change.

3. What are we doing to tackle AI biases?

From the algorithms that silently follow us around to the outputs we generate when we prompt - AI bias is influencing what we see. Human biases are well documented. By contrast, bias in AI algorithms often go undetected. This bias arises from training data that mirrors human prejudices or historical inequalities, even when sensitive variables like gender or race are excluded. AI can detect and replicate these biases through indirect patterns, highlighting the need for careful data scrutiny and algorithm refinement to ensure fairness. This is a problem, particularly when it comes to children’s usage in unsupervised environments.

When introducing AI bias to Year 7 students (11-year-olds), we begin by exploring where the data is trained, often tracing it back to Silicon Valley. From there, I challenge them to investigate who trains the data and the knowledge base it’s built upon - a task they quickly discover is more complex than it seems. Next, they ask ChatGPT to list the Top 10 scientists of all time. The response? 9 males and just 1 female, usually Marie Curie. The absence of other prominent female scientists leaves the students puzzled, sparking insightful discussions about representation and bias in AI systems.

Good work is happening to decolonise curriculums globally. In schools, children are coming into contact with new authors, ideas and cultural stories. This is progress we should be proud of. But when it comes to AI, we need to ask ourselves where the data is being trained, and by whom?

4. What epistemological framework does AI sit within?

When it comes to exploring the difference between knowledge and belief, AI presents us with a shifting epistemology. As a rule, AI generates outputs and insights without ‘understanding’. By identifying relationships in big data - LLM’s refine ‘what is out there’ in terms of knowledge, then draw a boundary around it. This is the output.

As a blanket term ‘Artificial Intelligence’ fits within a range of epistemological frameworks, including empiricism, constructivism and pragmatism. On a bad day, AI becomes intertwined with catastrophe theory and existential risk, reflecting the potential for small missteps to escalate into monumental disasters. Catastrophe theory highlights how gradual changes in a system can suddenly lead to drastic, irreversible shifts - much like an AI system gone rogue or the time Google told its users to eat rocks!

Another contender is post-humanism. The argument that AI could flourish in a post human world stems from the idea that it challenges the traditional notions of human-centred thought. When AI behaves well, meaning it functions effectively, ethically, and aligns with its intended purpose - it demonstrates the potential to transcend human limitations, complementing and reshaping what it means to ‘be.’ As a framework, post-humanism embraces the blending of human and non-human entities, encouraging a shift away from anthropocentrism (the belief that humans are the central or most significant entities).

As we move into another year of rapid innovation, it's clear that AI has become more than just a tool, it's a catalyst for questions that demand answers. From bias and environmental impact to fatigue and epistemology, the narratives we choose to amplify will shape how AI integrates into our world. It’s time to step beyond the noise, confront the silences, and uncover the stories that need to be told. The future of AI isn’t just about technology. It is about us. Let’s make sure we’re asking the right questions.

References, and thanks

Ideas and research for this post came in part through conversations with the Oxford University AIEOU community. I would also like to acknowledge Sara Watt for her guidance and knowledge around the AI and sustainability research.

Business Today ‘Microsoft water usage’ Link

Harvard Business Review ‘What do we know about the biases in ai? Link

WWF ‘AI and conservation’ Link:

Thomas Rudoff ‘Revealing the complexity of Fatigue’ Brain Sciences. Vol.14