AI can build the student startup. That may make entrepreneurship harder to teach

Generative AI is making it easy for students to produce polished products before they have established that anyone wants them, argues entrepreneurship educator Steve Blank. His warning is not that students can build too much, but that a convincing demo can now arrive before the messy learning begins.

Watch Desk analysis

What happened

In a post published on 23 September, Blank argues that generative AI lets students create complete products and apps almost immediately. He says this risks replacing customer discovery with what he calls “evidence theatre”, where a finished-looking product is mistaken for evidence of customer value.

Blank also argues that AI can intensify confirmation bias. Students may become attached to ideas that have not been properly tested, then overlook negative feedback because the software has made the idea feel real before customers have had a meaningful say.

Why it matters

The traditional minimum viable product was supposed to be a learning device, not merely a small product launch. If AI can generate the product in a day, the difficult work shifts elsewhere: finding the right customers, asking questions that do not lead the witness, and accepting that a neat solution may solve nobody’s problem.

That matters well beyond business-school classrooms. The same pattern applies to workplace prototypes and AI-assisted software projects, where speed can create an impressive artefact without producing reliable evidence. A machine that removes the early friction may also remove some of the useful doubt. Progress, it turns out, can be suspiciously well dressed.

Our read

Blank’s argument is an informed warning, not proof that every AI-built product is empty or that the MVP is obsolete in every setting. But the central point is strong: educators and teams should assess the quality of the learning and customer evidence, not reward the existence of a functioning app.

The sensible response is not to ban generative AI. It is to make the unglamorous work harder to fake. Require records of customer conversations, rejected assumptions, changes made after disconfirming feedback and evidence that comes from people outside the builder’s immediate circle.

What to watch

• Assessment criteria:
  whether entrepreneurship courses begin grading discovery and evidence rather than polished prototypes.
• Customer evidence:
  whether AI-assisted teams can show independent feedback that changed the product.
• MVP practice:
  whether “minimum viable product” evolves into a test of learning rather than a race to ship.
• Founder psychology:
  whether rapid construction makes people more willing to abandon weak ideas, or more invested in defending them.

Discussion spark: If AI makes a convincing prototype cheap, should entrepreneurship courses stop grading the product and grade the quality of customer discovery instead?

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.