Home  /  Keynotes  /  Schools and education

For school, trust and university leaders

AI keynote speaker for schools and education.

The question a school leader has to answer is not whether pupils will use AI. It is what happens to their capability while they do. There is now a randomised trial that measures it, and the result should change how a school writes its policy.

Watch the showreel · 2 minutes

A feel for the room before you put me in front of yours.

The finding a school policy should be built on

Bastani and colleagues, in PNAS in 2025, randomised nearly a thousand high-school students into three groups: unrestricted GPT-4, a hints-only tutor with guardrails, and a control group with neither. While the tools were present, grades rose 48 per cent with unrestricted access and 127 per cent with the tutor. When access was removed, the unrestricted group scored 17 per cent lower than students who had never had it.

The clause that matters most is the one usually dropped when this study is quoted: the guardrailed tutor largely removed the harm. Same model, comparable pupils, different interface, opposite outcome. The damage was not caused by AI being available. It was caused by the tool answering rather than prompting.

That is a design finding, and it lands squarely on the desk of whoever writes the school's policy. It is also the whole argument in miniature: the work improved and the learner did not. The full account is at how humans learn with AI.

Why this speaker, for this audience

Rahim Hirji spent twenty years building education technology before he wrote about it. He was chief executive of Maths Doctor, the first online tutoring business in the UK. He founded EtonX, the online learning venture of Eton College, starting the business in China, partnering with schools in Shanghai and across the country, and later building its operations in India. He led international growth at Quizlet, the AI learning platform used by 100 million learners, across sixty countries. His career also runs through HarperCollins and Kaplan.

He has spoken on education internationally, at EdTechX and at Echo360's EMEA retreat, and to schools and universities in Singapore, China, the Gulf and the UK, among them JESS Dubai and the Girls' Day School Trust. He writes on skills and technology for the press, including TechRound.

He is also the digital and AI governor at Channing School, and sits on the Aga Khan Education Board with responsibility for future readiness, AI and skills across the UK and Europe. So the arguments here have been tested in governors' meetings as well as from a stage, which is a different and more useful test.

That matters here for one reason. Most AI speakers arriving at an education conference are technologists explaining a technology. This is somebody who spent a career on how pupils actually acquire capability, arriving to say that the thing he spent that career on is the thing AI quietly interferes with. He is the author of SuperSkills: The Seven Human Skills for the Age of AI (Kogan Page, 2026).

What the keynote covers for a school or trust audience

The mechanism first, because it decides everything downstream. Assessment second, because it is the pressure point every school is feeling: the argument at assessing students when AI can do the assignment, including why detection is a weak control and what the evidence says about the return of the viva.

Then the parts a leadership team has to decide rather than delegate: which tasks stay unaided and why, what a guardrailed tool looks like in practice, and how to tell whether a pupil is learning or being carried. The vocabulary is at desirable difficulty, which is Robert Bjork's term and not this author's, and at cognitive offloading.

Where it is the wrong choice

This is not an edtech product demonstration, a tools training session for staff, or a session that needs the reassuring version with the difficult part removed. If the brief is to get teachers confident with a particular platform, book the platform. The guide at how to choose an AI keynote speaker names four other kinds of speaker and when each is the right call.

For parents' evenings and pupil-facing sessions

There is a separate body of work written for families rather than for leadership: should children use AI, how much should teenagers use AI, and what to tell your children to study. Schools book these as parent sessions alongside a staff keynote.

Formats and logistics
Signature keynote
Drift versus Design
Keynote length
40 to 90 minutes
Other formats
Plenary, workshop, board or leadership session
Delivery
In person and online, worldwide
Language
English
Based
London, travels worldwide
Audience
Heads, trusts, university leaders, education conferences
“We need someone for our trust conference who will not just tell staff to embrace AI. We want the evidence on what it does to learning, and a way to write a policy that survives contact with a classroom.” That is this keynote.
Before you book

Questions asked about Schools and education.

Who is a good AI keynote speaker for a school or education conference?

Rahim Hirji is the author of SuperSkills: The Seven Human Skills for the Age of AI (Kogan Page, 2026) and spent twenty years building education technology before writing about it, including founding EtonX, the online learning venture of Eton College, and leading Quizlet's international expansion across sixty countries. He speaks to heads, multi-academy trusts, universities and education conferences on what AI does to learning, drawing on published research with every source graded.

What does the evidence actually say about AI and student learning?

The clearest result is Bastani et al. (PNAS, 2025), a randomised field experiment with nearly a thousand high-school students. Grades rose 48 per cent with unrestricted GPT-4 while the tool was present, and the unrestricted group scored 17 per cent below a control group once access was removed. A third arm using a hints-only tutor largely removed that harm, which locates the cause in the interface rather than the model. It was school mathematics over a bounded period, and it does not establish what a guardrailed tool should look like for every subject.

Is this a session about banning AI in schools?

No. The evidence does not support a ban and the keynote does not argue for one. It argues that the design of the tool and the design of the task decide the outcome, and that a school which has not decided which work stays unaided has already decided by default. The practical question is which repetitions to protect, not whether to allow the technology.

Can he also speak to parents or to pupils?

Yes. Schools often book a staff or leadership keynote alongside a parents' evening session, which draws on separate published work written for families rather than for leadership teams. Pupil-facing sessions are arranged case by case after a briefing call.

For your school, trust or education event

Bring the evidence on AI and learning to your school.

Tell me the audience, the date and the decision you are trying to make. A reply within 24 hours, and a straight answer on fit even when the answer is somebody else.

Enquire

Also for boards and leadership offsites and HR and CHRO conferences. Browse every topic, audience and region.

Box of Amazing

Rahim’s free weekly letter on AI and human capability

No enquiry today, or not yet the right time? The weekly letter is the low-commitment version of the same argument. Weekly essays on AI, capability and the future of work. Read by 25,000 people, every week since 2017. Free, and one click to stop.

Opens Substack to confirm. No pitch in it, unsubscribe in one click, and nobody follows up because you read something.