The days when a lecturer could stand at the front of a room, talk for an hour and expect students to remain attentive throughout are gone. The conditions in which students learn have changed. They now study amid digital interruption, with information that is searchable, easy to revisit and available on demand. Simply speaking at them no longer fits how many engage with knowledge.
This shift had begun long before generative artificial intelligence appeared. Online journals, recorded teaching and educational videos already made expert explanations available beyond the lecture theatre. GenAI changes this relationship more fundamentally. Students no longer have to search across sources and piece an explanation together. AI can retrieve, synthesise and tailor information in response to a question, then refine its answer immediately. In 2026, 95 per cent of UK undergraduates reported using AI in at least one way, while 94 per cent said they used GenAI to help with assessed work. A student can request an explanation, worked example or comparison of theories within seconds, then ask for a simpler version or another example.
- Harness the power of ‘what if?’ in STEM teaching
- The Democracy Lab: how 20 minutes of debate can revive the lecture
- The lecture is dead – long live the performance
This does not make lecturers redundant; it makes one-way content delivery a weaker reason to bring students together. Expert explanation, intellectual storytelling and the modelling of disciplinary thinking still matter, but in-person teaching must also help students apply knowledge, question claims, build arguments and encounter perspectives beyond their own.
The problem is not lecturing; it is speaking for an hour and treating content coverage as evidence that learning has occurred.
These four practical shifts can make lectures more valuable in the age of AI.
1. Start with the thinking, not the slides
Academics often plan a session by asking: what content must I cover? This usually produces a list of topics and a growing slide deck. Now that students can ask a chatbot to define a concept, summarise a theory or generate an example, that is a weaker starting point.
A better question is: what should students be able to do with this knowledge? They might need to distinguish correlation from causation, select an appropriate analysis, evaluate an AI-generated explanation or apply a theory to an unfamiliar case. Once that task is clear, it becomes easier to decide what needs explaining, demonstrating and practising.
Content still matters, but the aim is not to get through as much material as possible. It is to choose and sequence material that helps students understand, apply, question and evaluate it. That was good practice before AI. What AI changes is the structure of the session: students can generate basic explanations and examples for themselves, so more lecture time can be organised around applying ideas, comparing interpretations, challenging claims and evaluating the quality of answers.
2. Interrupt fluency with retrieval and application
AI can produce clear explanations, polished summaries and plausible answers within seconds, but students still need to interrogate and digest that output. Reading a convincing explanation is not the same as being able to recall, explain or apply the idea independently. The same is true of a lecturer’s explanation; students may follow it in the moment without being able to use the knowledge themselves. A large meta-analysis of undergraduate STEM teaching found that active learning improved performance and reduced failure compared with conventional lecturing.
After explaining a concept, pose a focused question. Give students 10 seconds to think and commit to a response, then 20 seconds to compare their reasoning with the person beside them, if they wish. They might predict the result of a study, identify the flaw in an AI-generated argument, select an analysis for a short scenario or apply a theory to a new case.
Longer explanations can also include a brief processing pause: 30 seconds to review notes, write down the main point, identify a question or clarify an idea with someone nearby. The point is to give students time to digest and use knowledge themselves before another answer – whether from the lecturer or AI – is supplied.
3. Design participation without demanding performance
Education cannot be reduced to a private exchange between a student and a chatbot. Yet active teaching can become compulsory public performance. Anxiety, disability, language background and personal preference all affect how comfortable students feel speaking aloud in a crowded lecture theatre.
So, offer different ways to contribute. Students might think privately before comparing an answer with one person, respond anonymously through Padlet or a polling tool, or write a short explanation. Use the final five minutes for a multiple-choice poll or an exit ticket asking students to identify one point they understand and one that remains unclear. Common questions can then be addressed in the next session or through the virtual learning environment.
Participation does not have to be loud. What matters is whether students are testing ideas, encountering other perspectives and moving beyond the first plausible answer AI provides.
4. Teach judgement, not just answers
AI can fabricate references, flatten disagreement and present weak claims with impressive confidence. Students therefore need to see how experts judge an answer: what evidence supports it, what assumptions it makes, what has been omitted and whether the conclusion goes beyond the data.
Bring an AI-generated response, media claim or flawed interpretation into the room and examine it together. Students might compare two AI responses, trace a claim back to its source or rewrite an overconfident conclusion so that it matches the evidence. The lecturer’s value lies in making expert judgement visible. Students need to see how knowledge is questioned, verified and revised.
This is not a demand for lecturers to become entertainers or to turn every explanation into an activity. Students still need sustained explanation, complex arguments and expert guidance. But an hour of shared human presence should offer more than information delivery. If a session can be replaced without loss by slides, a recording or an AI summary, students will notice.
GenAI has not destroyed university teaching. It has exposed a weakness that was already present: information delivery was never the same thing as education. Students can obtain information almost anywhere. What they still need from university is help learning how to think with it.
Debra Page is a lecturer in the School of Psychology and Clinical Language Sciences at the University of Reading, UK.
If you would like advice and insight from academics and university staff delivered direct to your inbox each week, sign up for the Campus newsletter.
comment