AI & The Next Phase - Cottesmore School

Thank you to Tom Rogerson and Cottesmore School for hosting such a thoughtful and challenging day on artificial intelligence in education. The event created space for honest and provocative conversations, treating AI not as a passing trend but as a structural shift that demands deep reflection. The speakers prompted us to question assumptions about assessment, inclusion, ethics and agency, and what emerged was not a single narrative but a shared recognition of urgency. AI is no longer on the horizon; it is already woven into our systems and our children’s lives, and schools must respond with wisdom and care.

Sir Anthony Seldon

Anthony Seldon opened with a clear warning. AI, he argued, is advancing faster than our capacity to civilise or humanise it. He placed this moment in a historical frame, comparing its impact to the invention of the printing press, and called it “the biggest thing to come into schools ever”. His provocation was not about technology but about the moral and human questions it raises. What should education plant in young people that will last a lifetime?

Seldon spoke about the erosion of empathy, curiosity and agency, qualities already under threat in a culture shaped by assessment and metrics. He warned that AI could accelerate this narrowing if education does not consciously protect its moral purpose. His comment that “AI is not the end or purpose of education, it is the means to the end” underscored a central theme of the day: technology cannot lead the agenda.

His vision of education was deeply humanist. He pointed to Howard Gardner’s line, “Do not ask how intelligent a student is; ask how a student is intelligent”, as a challenge to deficit-based thinking and a reminder that intelligence is diverse and relational. This diversity, he argued, will be harder to defend in an era where technology reduces children to data points. Seldon’s message was clear: educators must act collectively, rather than waiting for governments or corporations, to ensure AI strengthens rather than erodes what it means to be human.

Rose Luckin

Rose Luckin’s session provided a rigorous framework for understanding AI. She reminded us that ChatGPT’s sudden ubiquity is not evidence of a technological revolution so much as a turning point of accessibility. Its 100 million users in two months reflect a perfect storm of computing power, data and algorithms that has made AI tools available to the general public. Yet she warned against assuming that accessibility equates to capability.

Luckin argued that schools need a more precise understanding of intelligence. She differentiated between human intelligence, which is experiential and embodied, and AI, which learns only through mediated data. Machines can approximate human reasoning but cannot access tacit or physical knowledge. She offered a taxonomy of intelligence, including academic, metacognitive, social, contextual and subjective forms, and emphasised the importance of self-efficacy. Her conclusion was that education must aim not only to preserve but to deepen these qualities in an age where machine intelligence is easy to mistake for human understanding.

She also warned of worrying trends. Research has shown a decline in reasoning and problem-solving ability since 2012, a trend that has coincided with increasing reliance on digital systems. Students are already deeply embedded in this technological shift, with 92% reporting that they use AI in their studies. Luckin’s 4D AI Strategy Framework offered a practical response: governance and ethics must lead, a clear vision must be articulated, staff capability must be developed, and data and technology should be used to support rather than dictate decisions. Her analysis set the tone for the day: AI’s power is real, but so are its limitations, and schools must understand both.

Chris Goodall

Chris Goodall brought a grounded perspective, challenging the current pace of AI adoption. He described schools as being caught in a “gold rush”, rushing to implement tools without pausing to ask why. His analogy was striking: AI is not lightning but electricity. Its power lies in infrastructure, not spectacle, and this makes thoughtful planning more essential.

Goodall argued that schools should begin with problems, not products. Examples such as supporting enquiry-based learning, strengthening literacy, or improving parent communication illustrated how AI can be used to address genuine needs. Yet he cautioned that these opportunities will be undermined if technology is introduced without clarity of purpose.

Equity was a recurring theme in his talk. Goodall highlighted that AI acts as an amplifier, widening existing gaps in access and outcomes. He expressed concern that detection tools are disproportionately penalising vulnerable students, and warned that unless this is addressed, the technology will entrench rather than reduce inequality. His starkest observation was that “we are automating an already broken system”. AI’s introduction is not inherently progressive; its impact will depend entirely on whether educators use it to repair or to replicate flawed structures.

Edufuturists

Ben Whitaker and Steven Hope’s session focused on the skills gap between what schools deliver and what society increasingly demands. With 83% of companies citing AI as a top business priority, they argued that education cannot afford to treat technology as a separate subject. Instead, they called for a pedagogy that prioritises agency, critical thinking and ethical decision-making.

The pair rejected binary debates about banning or embracing devices. Instead, they advocated for “guardrails” — structures that scaffold children’s autonomy and responsibility. Rather than restricting access, they argued that schools must teach children to manage distraction and navigate digital systems critically. This approach reframes safeguarding not as a set of prohibitions but as a preparation for life.

Their reflections on assessment echoed others throughout the day. They argued that exams, while still useful, are no longer adequate as a single measure of learning. In a world of rapid technological change, assessment should value process, dialogue and creativity as much as outcome. This requires a cultural shift, one that sees technology not as a threat but as a reason to re-examine the deeper purposes of education.

Laura Knight

Laura Knight’s Wired for Wisdom session offered one of the most critical perspectives of the day. She argued that schools risk repeating the mistakes made with social media adoption, rushing into procurement decisions without anticipating long-term consequences. She warned that some choices made now will “look ugly” in retrospect.

Knight’s focus was on sovereignty: the ability of children and teachers to own their data, identities and decisions in a digital world. She emphasised that safeguarding must go beyond external controls, building internal moral compasses in young people. She pointed to a growing “chasm” between families and schools, with parents often disengaged and children carrying shame or secrecy in their online lives.

Her reflections on learning were particularly resonant: “If learning feels easy, it’s not real learning.” She argued that Big Tech’s drive to remove friction risks eroding resilience and criticality, and that education must reclaim discomfort as a sign of growth. Knight linked these issues to democracy itself, warning that schools are unprepared for the social consequences of AI-driven manipulation and disinformation. Her vision was not anti-technology but deeply ethical: children must learn not only to use AI but to interrogate it.

Patrick McGrath

Patrick McGrath’s talk reframed AI as a tool for equity, not simply efficiency. He argued that discussions about accessibility are too narrow and should encompass mental health, diversity, masking and the invisible challenges students face daily. He proposed Universal Design for Learning as a model for this work, advocating for multiple means of engagement, expression and access.

McGrath’s examples illustrated AI’s capacity to support learners in deeply human ways: multimodal tools for dyslexic students, podcasts for parents, and even the recreation of a lost voice. Yet he also drew attention to stark inequalities. Only 9% of state-maintained schools, 3% of special schools and 22% of independent schools currently have a clear AI policy. Without systemic planning, he argued, AI’s benefits will be reserved for those already advantaged. His emphasis on dialogue, student agency and personalised learning offered a vision of AI as a leveller, but only if deployed with care.

Dan Fitzpatrick

Dan Fitzpatrick closed the day with a future-facing perspective. He predicted that robotics and neural interfaces may represent the next wave of disruption, perhaps as significant as ChatGPT’s rise. Children leaving school may enter a world where technologies like wearable devices, thought-to-text systems and domestic robots are commonplace.

Fitzpatrick’s focus was not just on tools but on readiness. He argued that current assessment systems are misaligned with this reality, treating assignments as logistical hurdles rather than meaningful intellectual work. He highlighted the risk of disengagement: students are not motivated by compliance, and AI will only deepen that disconnection unless education becomes genuinely relevant.

He asked a crucial question: if enterprise owns production, how should education evolve? His live demonstration of an AI-generated business plan underscored how easily students can outsource surface-level tasks. The challenge for educators, he concluded, is to design learning experiences that cannot be reduced to outputs, cultivating curiosity, criticality and ethical judgement.

Closing Reflections

The conference revealed not a single narrative about AI but a complex and sometimes uneasy picture. Across all sessions, speakers resisted simple optimism or fear. AI was framed as a mirror, amplifying the systems and priorities already in place. If education is reductionist, AI will accelerate that reduction. If it is relational, ethical and critical, AI may extend its reach.

Several questions linger. How do schools protect the moral purpose of education while embracing technological tools? What systems of assessment are adequate for a generation that can automate content creation? How can educators lead rather than follow when policy and procurement decisions are being driven externally? And perhaps most importantly, how can schools ensure that AI amplifies equity rather than inequality, giving every child the right not only to use AI but to understand and question it?

Thank you again to Tom Rogerson and Cottesmore school for such a valuable and thought-provoking day!