Give the classroom a job the explanation cannot do alone

TL;DR: I would give classroom time a visible job: let students expose a piece of work to questions, then revise it. This classroom discussion shared as Pratosh’s IISc lecture is useful for instructors and team leads redesigning learning around AI. Its strongest prompt is about what people should do together when explanations are readily available.

The conversation moves through attendance, motivation and the difficulty of assessing work that a model can produce. Later, it considers small teams building something for users, while acknowledging that simple popularity measures can be gamed. That unresolved discussion is more useful to me than a prediction that teachers will disappear.

An explanation can help a learner understand a method. It does not by itself establish that the learner can choose the method appropriately, defend the choice or respond when somebody finds a flaw. Shared time can be organised around those missing activities.

Make the artifact carry the discussion

Imagine a learning session about using AI to organise a help desk. I would ask each small team to bring a working example, the requests it is meant to handle and one request it fails on. This is a proposed class exercise, not a study result.

Another team would act as the receiving user. They would try to complete the intended task, describe what confused them and ask why the system behaved as it did. The presenting team would have to separate a design choice from a limitation it had not previously noticed.

The instructor’s job becomes concrete: ask for the evidence behind the explanation, surface a relevant concept when it is missing, and keep the critique focused on the work. A polished answer that cannot account for the artifact should not end the discussion.

Then reserve time for a revision. Students should show what they changed and why. Without that return step, the critique risks becoming another performance that ends when the session ends.

Assess the learning the project makes visible

I would not grade this exercise simply by revenue, stars or the number of people a team can recruit. Those measures can reflect access, promotion and timing as much as understanding. A project with few users can still demonstrate careful reasoning and a meaningful correction.

A useful review would examine whether the team identified a real need, used evidence to choose an approach and could explain a revision. For work involving an outside user, the exercise must also respect that person’s time and permission. A classroom does not need to create a company to produce useful feedback.

The source is an exploratory conversation. It is not evidence for its broader claims about the future economy, consciousness or the end of intellectual work. Nor does the proposed exercise replace foundational teaching or individual assessment. Students still need opportunities to demonstrate what they understand themselves.

My bet is narrower: a class earns shared time when another person’s question can improve the work before everyone leaves. The teacher remains essential to making that exchange rigorous and useful.

Use the classroom for work that becomes better through examination and revision in front of other people.