Matt and Al are joined by Daniel Hulme, one of the UK's leading voices on AI. Daniel grew up in Morecambe, was actively discouraged from applying to university, and did his A levels under a desk in the hotel restaurant he was running at 16. He got into UCL anyway, fell in love with education, and went on to found the AI company Satalia (acquired by WPP, where he is now Chief AI Officer) and Conscium, a company working on machine consciousness. He also ran a master's programme at UCL for five years.

The conversation starts with a blunt question: is the traditional school-to-university pathway still an honest transaction? Daniel's answer is nuanced. He's watching students opt straight into work rather than risk three years on a degree with no job guarantee, and he doesn't think they're wrong. But he's equally clear that foundational knowledge still matters: the fundamentals of algorithms and computational complexity haven't changed in decades, and they're what separate people who can make good decisions about AI from the crowd who rebranded themselves as experts three years ago.

He's forthright on the state of AI adoption too, borrowing Dan Ariely's line about big data: everyone says they're doing AI agents, almost nobody is, and much of what the big consultancies are selling is nonsense. His advice is the same one educators give: start with the problem, not the technology.

From there it gets properly mind-stretching: why large language models are 'intoxicated graduates' on their way to professor level, spiking neural networks and brains that run on 20 watts, evolving digital creatures in a virtual petri dish, his colour-wheel theory of consciousness, the seven singularities, and whether a machine that understands pain and empathy would be safer for humanity than a 'zombie superintelligence'.

The closing stretch comes back to education: testing how students apply knowledge rather than whether they hold it, AI's potential to narrow (or widen) disadvantage gaps, and what people say they'd do in a world of abundance. Top answer: improve their golf handicap. Second answer: learn.

Daniel's new book, Perspectives on Machine Consciousness, gathering 30 of the world's leading thinkers in the field, is out in the coming months.

Matt and Al are joined by Mike Berners-Lee, professor at Lancaster University, founder of Small World Consulting and author of four books including How Bad Are Bananas? and There Is No Planet B. Mike chairs the National Emergency Briefing, which last November brought 1,200 of the UK's most influential people, including around 85 MPs, peers, business leaders and faith leaders, into Central Hall Westminster to hear nine top experts set out the bare facts on the climate and nature emergency. The government should be briefing the public the way it did during COVID. It hasn't, so they did.

Mike walks through what those nine briefings covered: CO2 levels now far outside the narrow range that supported 10,000 years of human civilisation, the rising risk of the AMOC ocean current shutting down and what that would mean for UK farming, weather extremes that our century-old infrastructure was never built for, food security, health, defence, and the economics of the transition, which contrary to the usual noise pays for itself by about 2052.

It isn't a counsel of despair. Not one person in that room of 1,200 thought the picture had been exaggerated, and the energy in the room went up as the dots were joined. The briefing has since become the People's Emergency Briefing, a 50-minute film presented by Chris Packham, now past 1,600 community screenings across the UK, with each screening triggering more than one new one on average.

Along the way: disinformation and the psychology of denial, whether the UK is too small to matter (it isn't), how to talk to children honestly about the crisis, critical thinking in schools, the Inner Development Goals, and how Mike stays positive after decades of banging the drum.

Take action at nebriefing.org: find or host a screening, check whether your MP attended the briefing, and sign the petition calling for a televised national emergency briefing across the major TV channels.

Could AI help school leaders understand Ofsted, identify hidden risks and approach inspection with greater confidence?

Victoria Hedlund—founder of GenEd Labs and widely known as “The AI Bias Girl”—returns to Joining the Dots to discuss Inspection Insights, her data-led approach to analysing inspection reports and supporting school leaders.

Victoria joins Al Kingsley and Matt Jessop to examine the human biases that can influence inspection. Every inspector brings personal values, experiences and assumptions to the process—but how effectively does the inspection framework recognise and mitigate them?

The conversation also explores whether judging every school against national averages creates structural bias. School size, phase, pupil demographics and socioeconomic context can all affect what success looks like, making a single national yardstick difficult to apply fairly.

Victoria explains how AI can analyse published reports, school performance information and demographic data to identify patterns, benchmark similar schools and generate reflective questions. The aim is not to predict or manipulate an inspection outcome, but to offer leaders another informed perspective and highlight issues they may not have considered.

In this episode:

Human and structural bias within inspection

Why context matters when evaluating schools

The particular pressures facing small schools

Analysing inspection reports with AI

Benchmarking schools against similar settings

The ethical risks of profiling individual inspectors

What inspection reports reveal about AI governance

The gap between national AI guidance and accountability

Why examples of effective AI practice should be celebrated

Combining inspection data, policies and school self-evaluation

Using AI as a reflective tool—not an infallible judge

Helping leaders feel informed rather than overwhelmed

Victoria describes the resulting analysis as a diagnostic snapshot: valuable information that can support reflection and planning, while never replacing professional judgement or the human elements of inspection.

Explore Inspection Insights:

Learn more about Victoria and GenEd Labs:

Explore Victoria’s services for school leaders:

Educational success rarely follows a perfectly straight line—and Dr Caroline Keep’s journey is powerful proof of that.

In this episode of Joining the Dots, Al Kingsley and Matt Jessop speak with Caroline about progressing from an underestimated, neurodivergent learner to an award-winning teacher, data scientist and EdTech researcher with a PhD.

Caroline reflects on being autistic, dyslexic and having ADHD, the low expectations placed upon her during childhood, and the transformative support she received from The Prince’s Trust. Her experiences now shape her determination to champion learners whose abilities may be overlooked by conventional education.

The conversation explores why traditional assessments often fail to capture what learners genuinely know and can do. Caroline explains how authentic assessment and more nuanced approaches to decision-making could help education move beyond rigid categories and arbitrary measures of success.

They also discuss Caroline’s use of fuzzy logic in educational research. Unlike binary models that reduce decisions to right or wrong, ready or not ready, fuzzy logic recognises degrees of difference—reflecting the complex reality of learners, classrooms and technology adoption.

In this episode:

Caroline’s journey from underdog to PhD

Autism, ADHD, dyslexia and educational expectations

Why every learner’s pathway looks different

The importance of authentic assessment

Moving beyond rigid measures of ability

How fuzzy logic can inform educational decisions

Rethinking the post-COVID language of “catch-up”

Accessibility through dictation, captions and digital tools

Why EdTech decisions should begin with educational values

Putting pedagogy before technology

Caroline’s position is neither pro-technology nor anti-technology. She is pro-pedagogy: choosing tools because they meet the needs of learners, not simply because they are new.

Learn more about Caroline:

Connect with Caroline on LinkedIn:

Digital transformation in education is not about placing a device in every learner’s hands. It is about leadership, trust, professional learning—and choosing the right tool for the right educational purpose.

In this episode of Joining the Dots, Al Kingsley speaks with Dr Rebecca “Becki” Bawler about digital competence in Wales and the lessons education systems can learn from Estonia’s carefully planned, pedagogy-first approach.

Becki explains why digital competence cannot be limited to learners. Teachers and leaders also need the confidence, skills and time to build digitally capable learning environments. Yet following the rapid adoption of technology during COVID-19, many schools have returned to familiar teaching practices rather than building on what they learned.

The conversation examines the role of leadership in creating a strong digital culture, the value of informal and community-based professional development, and why trust is essential when encouraging colleagues to change their practice.

Al and Becki also explore Estonia’s education system, where digital transformation has developed through careful planning and necessity rather than technology for technology’s sake. They consider what Wales can learn from this approach—and why Wales may already be further along that journey than it sometimes recognises.

In this episode:

Why digital competence must include teachers and leaders

Leadership’s role in creating a confident digital culture

What happened to digital adoption after COVID-19

Building trust through informal, community-based learning

Why Estonia plans before implementing technology

Using AI to strengthen critical thinking

Putting pedagogy before platforms and devices

The potential for hybrid learning

Why infrastructure remains a significant barrier

The logistical challenges surrounding digital examinations

Working with industry to create relevant curricula

Professional learning as a factor in teacher retention

What England and Wales can learn from one another

The central message is clear: meaningful digital transformation begins with the needs of the classroom. Technology should support good pedagogy—not dictate it.

Explore the complex relationship between AI and bias with expert Victoria Hedlund. Learn how AI perpetuates biases and what can be done to mitigate them.

AI promises greater personalisation, productivity and accessibility in education—but what happens when its answers quietly reproduce the biases already present in society?

In this episode of Joining the Dots, Al Kingsley and Matt Jessop speak with Victoria Hedlund, founder of GenEd Labs and better known as “The AI Bias Girl”, about the assumptions hidden inside generative AI.

Victoria reveals how factors such as a learner’s name, gender, postcode or perceived background can change the guidance AI provides. From university advice to lesson planning, these subtle differences could influence attainment, confidence and opportunity without teachers or students even noticing.

The conversation also explores the tension between personalisation and a shared curriculum. If every learner receives a completely different AI-generated explanation, could we lose the common language and conceptual foundations that enable people to learn and work together?

Victoria argues that educators must remain the “teacher in the loop”—questioning outputs, protecting human agency and treating every AI response as an experiment rather than an objective truth.

In this episode:

How hidden bias appears in AI-generated content

Why names and postcodes can influence AI responses

Whether completely unbiased AI is possible

The danger of AI-generated content training future models

Why children may be more experimental with AI than adults

The training and time teachers need to use AI responsibly

How over-personalisation could fragment the curriculum

Why teachers may become curriculum anchors and pathway managers

Practical questions that expose assumptions in AI outputs

Why adaptable AI strategy matters more than another static framework

Victoria’s practical advice is simple: put every AI output on trial. Ask what assumptions it has made, how it has affected your choices and what evidence supports its answer.

Learn more about Victoria and GenEd Labs:

Explore Victoria’s books and resources:

Discover the Institute of AI Education:

Al Kingsley and Matt Jessop discuss with Heather Rhodes insights on online education, its impact on soft skills, and innovative projects in India, emphasising personalised learning, hybrid models, and future educational reforms.

key topics

Building the Future of Education with AI and System Innovation

Navigating Education's Evolving Landscape: Insights from Andy Kent on Change and Data

In this episode, Matt Jessop and Al Kingsley delve into the profound shifts in education over the past 27 years, especially accelerated post-COVID, and how data and AI are shaping the future of teaching and school management. Join us for an enlightening discussion with EdTech entrepreneur and all round nice guy, Andy Kent, as we explore the challenges, opportunities, and innovations transforming education today.

Key Topics:

This episode offers valuable insights into how education professionals can leverage data, AI, and adaptable practices to navigate ongoing challenges and build resilient, innovative learning environments.

In this conversation, Caroline Keep discusses her work in AI, education, and community engagement with Al Kingsley and Matt Jessop. She emphasises the importance of accessibility and neurodiversity in educational technology. She shares insights from her research on fuzzy logic and the challenges of AI, as well as her role in shaping policies in Liverpool. Caroline also reflects on her experiences in teaching and community outreach, highlighting the need for more inclusive practices in education.

Takeaways