An AI keynote can fill a conference hall, attract sponsors and leave the audience buzzing. But what happens on Monday, when teachers return to their classrooms and school leaders resume decisions about assessment, safeguarding and professional learning?
The best AI in education keynote speaker is not simply the person who knows the newest tools. It is someone who understands how people learn, can distinguish evidence from enthusiasm, translates AI developments into your audience’s real decisions, and gives educators a credible next step. A memorable presentation is valuable; a presentation that changes the quality of the conversation is more valuable.
For school networks, education conferences, teacher development events and universities, selecting that speaker is a pedagogical decision as much as an event-planning decision.
First, decide what the keynote must change
“AI in education” is a subject, not an event outcome. An organizer who asks only for a talk about AI may receive an entertaining demonstration of applications, a warning about cheating, or a broad forecast of the future. Any of those may be appropriate—but they solve different problems.
Start by completing this sentence: After the keynote, our audience should be better able to _____.
For example:
A school leadership conference: distinguish responsible AI adoption from tool accumulation, then identify which decisions require a school-wide policy.
A teacher professional development day: redesign one learning task so that AI assists the learner without replacing the thinking the learner needs to practise.
A university faculty event: separate a polished AI-assisted submission from reliable evidence of student understanding.
A cross-sector education summit: debate the conditions under which AI can improve access and learning without weakening human agency.
These aims lead to different speaker briefs, examples and evaluation methods. They also tell you whether a keynote is the right format. A plenary can establish a shared question; a workshop is usually better for creating or testing actual lesson materials.
Arizona State University's guidance on choosing a keynote speaker emphasizes expertise closely aligned with the challenge at hand. Its concrete example is particularly relevant: for an AI-in-education event, an education expert may be the stronger opening choice than a technology expert, because the educational problem deserves to be understood before proposing a technological answer.
What the evidence says a responsible speaker should know
The evidence base on generative AI and learning is developing quickly. That is precisely why your speaker needs intellectual discipline.
The OECD Digital Education Outlook 2026 distinguishes between producing a better answer with AI and learning something that can be demonstrated independently. The report synthesizes emerging research showing that general-purpose generative AI can improve task performance without necessarily producing corresponding learning gains. Purposefully designed educational uses—such as feedback, questioning or guided tutoring—may support learning more effectively. It also notes the importance of teacher agency and thoughtful pedagogy.
That distinction changes what a good keynote should demonstrate. A speaker could show a chatbot solving a science question in seconds. The more consequential demonstration would ask: Can a student explain the mechanism afterward? Can they detect an error? Can they use the idea in a new context without the tool?
UNESCO's AI Competency Framework for Teachers offers another useful benchmark. Published in 2024, it organizes 15 teacher competencies across five dimensions: a human-centred mindset, ethics of AI, AI foundations and applications, AI pedagogy, and AI for professional learning. A presentation concerned only with prompts and apps addresses a narrow slice of what professional competence requires.
UNESCO's guidance on generative AI in education and research also raises issues of privacy, age-appropriate use, inclusion and human oversight. A responsible keynote will not imply that a compelling classroom demonstration is sufficient proof that a tool is safe, compliant with local requirements or pedagogically effective.
None of these reports provides a universal recipe for hiring a speaker. They do, however, establish the issues on which an education audience should expect informed, cautious and practical discussion.
Seven criteria for evaluating an AI in education keynote speaker
1. Education expertise comes before tool fluency
Ask whether the candidate can explain attention, prior knowledge, feedback, assessment, learner independence and teacher judgment—not just how to generate a lesson plan. Someone who understands educational trade-offs will know when AI adds value and when a simpler human-led approach is better.
Listen for: a specific example that starts with a learning objective and ends with observable student work.
Be cautious when: every challenge is answered by recommending a new platform.
2. Evidence is presented with its limits
A strong speaker can say what a study actually measured, distinguish immediate task success from retained learning, and acknowledge uncertainty. They do not need to turn a keynote into a literature review. They do need to avoid the unqualified promise that AI “proves” higher achievement or that every use damages independent thinking.
Listen for: citations to identifiable research and an explanation of where the findings may not generalize.
Be cautious when: precise percentages, sweeping forecasts or personal anecdotes substitute for verifiable evidence.
3. The talk reflects your participants' daily reality
A school principal, a primary teacher, a university lecturer and an education minister face different choices. The examples should fit their learners, curricula, policies, infrastructure and responsibilities.
Listen for: questions about age groups, institutional context, country, audience experience, accessibility and whether the event is in person or virtual.
Be cautious when: the same slide deck is offered to every audience without adaptation.
4. Student agency remains central
AI can accelerate drafts, explanations and feedback. Your audience also needs to ask which cognitive work students must still own. A useful speaker should show how to preserve the learner’s role in questioning, verifying, reflecting and transferring knowledge.
For a classroom-ready example of this distinction, see Learning Coach's published article When AI Gives the Right Answer: How Can Teachers Know Students Really Understand?. Its central lesson is relevant to any keynote: a polished product is not the same as evidence of understanding.
Listen for: examples of students explaining and challenging AI outputs—not merely copying them.
Be cautious when: “personalization” is discussed without clarifying what the student learns to do independently.
5. Ethics and governance are treated as practical responsibilities
Generative AI raises questions about personal data, age-appropriate access, copyrighted material, bias, academic integrity, accessibility and accountability. Laws and institutional policies vary by jurisdiction, so speakers should avoid presenting one global rulebook as universally binding.
Listen for: a clear distinction between what teachers can decide, what requires institutional approval and what must remain a human responsibility.
Be cautious when: student data is casually entered into public tools during a live demonstration.
6. Stage presence supports substance
A persuasive keynote needs narrative, clarity, pacing and audience engagement. Ask for a complete or substantial sample of the speaker's work when available, rather than relying only on a polished highlight reel. References and previous appearances should be verifiable; expertise may also be evidenced through publications and professional work.
Event-planning advice from Cvent similarly stresses defining objectives, understanding the audience and reviewing the speaker's delivery before committing.
Listen for: a meaningful audience question, an intelligible example and a memorable idea that can be tested afterward.
Be cautious when: entertaining visuals obscure the absence of a clear argument.
7. The event has an honest pathway to follow-through
No speaker can credibly guarantee that a single hour will transform a school system. They can help design a session that starts a useful process: a question leaders will revisit, a shared vocabulary or a practical activity for participants to try.
Listen for: realistic outputs, an optional resource list, and clarity about what happens after the keynote.
Be cautious when: long-term impact is promised without follow-up, implementation support or an evaluation plan.
A comparison grid your selection committee can actually use
Use this table to structure a conversation—not to turn a subjective judgment into a scientific score. Compare candidates on the same evidence and ask each committee member to record uncertainties.
| Selection dimension | Evidence to request | Warning sign |
|---|---|---|
| Learning-science fit | Example linking an AI use to a specific learning goal | Only a software demonstration |
| Evidence quality | Current sources plus acknowledged limitations | Unverifiable statistics or universal claims |
| Audience fit | Audience discovery questions and adapted outline | One generic talk for every sector |
| Safety and ethics | Clear treatment of privacy, equity and human judgment | Risks treated as an afterthought |
| Delivery | Substantial sample, transparent experience, suitable format | Only slogans or highlight clips |
| Follow-through | Practical question or next action for participants | Claims of guaranteed organizational change |
A committee can mark each row as strong evidence, partial evidence, insufficient evidence or not yet assessed. If candidates are close, give greater weight to the dimensions that serve your stated event outcome. For example, a day intended to redesign assessments should prioritize learning evidence and teacher practicality over technological spectacle.
Ten questions to ask before you book
Send these questions to shortlisted speakers and use the answers in a short conversation:
What is the central question our participants will still be discussing a week after your talk?
Which parts of your keynote are grounded in published research, and where is the evidence unsettled?
How will you explain the difference between task performance and genuine learning?
Which examples will you adapt for our specific age groups, disciplines and country?
How will you address privacy, equity, academic integrity and institutional responsibility?
What could participants reasonably try or discuss the following week?
Can we see a substantial talk sample or other credible evidence of your communication style?
What do you need to learn about our audience before drafting the session?
Would a keynote, interactive conference or workshop serve our objective best—and why?
What exactly is included in the proposal: preparation, format, travel, recording permissions, materials and follow-up?
The point is not to catch a candidate out. It is to discover whether they can think with your organizing team rather than deliver a prepackaged performance.
Build the brief around a real decision
A useful speaker brief can be surprisingly short. Here is an illustrative example for an English-language school-network event:
Audience: 180 school leaders and teachers from primary and secondary schools, with mixed experience of generative AI.
Context: Several schools permit AI-assisted planning, but teachers are uncertain about classroom tasks and evidence of student understanding.
Key question: How can we use AI without teaching students to outsource the thinking we want them to develop?
Desired outcome: Participants agree on two principles for responsible use and each department identifies one task or assessment routine to examine next.
Possible structure for a 50-minute plenary: 10 minutes to frame the learning dilemma; 15 minutes to examine evidence and contrasting examples; 15 minutes for an audience decision exercise; 10 minutes for questions and next steps.
Boundary: This is a keynote, not a compliance audit, teacher certification or a guarantee of new student outcomes.
A well-prepared speaker might suggest changes. That is a positive sign: the organizer brings the context, and the specialist helps make the event more educationally coherent.
Clarify the commercial and operational details
Before confirming a booking, request a written proposal that states the session title and learning aims, duration, delivery format, audience size, languages, preparation requirements and fee structure. Clarify travel, accommodation and cancellation terms where relevant. Ask who owns presentation materials, whether recording or redistribution is permitted, and whether any product promotion will occur.
An international event also needs practical checks: time-zone alignment for virtual appearances, accessibility provisions, local policy sensitivities, on-site or online audiovisual requirements, and enough lead time to adapt examples. Never assume these are included unless the proposal says so.
Measure more than applause
Post-event satisfaction matters, but it is not the same as educational impact. A simple, proportionate evaluation can work at three levels:
Immediately: Ask participants to name one idea they would challenge or apply and one question that remains unresolved.
After two to four weeks: Ask whether teams held the planned discussion or piloted a suitable classroom or policy change. Record actual actions, not intentions alone.
Where feasible: Review a small sample of the resulting tasks, decisions or learning evidence against the original objective. Do not attribute every subsequent improvement to the keynote.
If the keynote opens a productive conversation but follow-through is weak, that is useful information for the next event. The solution may be a clearer brief, more time for discussion or an accompanying workshop—not necessarily a different celebrity speaker.
Frequently asked questions
Should an AI in education keynote speaker be a technologist or an educator?
It depends on the goal. A technical specialist may be ideal for explaining model architecture or infrastructure. For a conference focused on pedagogy, classroom practice and student learning, educational expertise should carry significant weight. Some speakers bring both perspectives; ask for evidence rather than relying on a job title.
Can one keynote cover teachers, school leaders and parents?
Yes, if the shared question is strong and the content is carefully adapted. But a single session cannot address every group's operational needs in depth. Consider one plenary followed by audience-specific discussions.
Is a keynote enough for teacher professional development?
Usually not when the intended outcome is a changed classroom routine. A keynote can establish purpose and a common language; practice, feedback and institutional follow-up are often needed to convert that into sustained professional learning.
The question worth putting on stage
Choosing an AI in education keynote speaker is ultimately a test of what your event believes education is for. If the aim is to display the speed of the latest technology, almost any impressive demonstration will do. If the aim is to help educators protect and deepen human learning, the standard must be higher.
The better question is not “Who can show us what AI can do?” It is “Who can help our community decide what learners should still be able to do for themselves?”
Bring that question to your next education event. Explore Learning Coach's international keynotes and request a tailored proposal for school leaders, teachers, institutions or university audiences.
References
OECD (2026). OECD Digital Education Outlook 2026: Exploring Effective Uses of Generative AI in Education.
UNESCO (2024). AI Competency Framework for Teachers.
UNESCO (2023). Guidance for Generative AI in Education and Research.
Arizona State University, University College. Selecting Your Keynote Speaker.
Cvent (2024). How to Secure Great Speakers for Your Next Event.
