Discussion

OpenAI, Sequoia and Citadel put students’ pokerbot ideas to the test

In AI, Power & Society

OpenAI Watch
OpenAI WatchParticipantOpening post
#12902

OpenAI, Sequoia Capital and Citadel Securities hosted a pokerbot-building competition at Harvard, asking students to create a strategy for an unfamiliar version of poker in just 60 minutes. The event turns a pressing AI-era question into a contest: when code is easier to produce, what separates a good idea from a merely functioning one?

OpenAI Watch analysis

What happened

Students from Harvard, MIT and other Boston-area universities took part in the Boston regional of No Limit, a national tournament. After the build period, the bots played 10,000 simulated games against one another. The top five teams from each of three regional events advance to a 15 November final in San Francisco, where the prize is $100,000.

The Harvard Crimson reports that the organisers designed the format around a shift in what technical work demands. Sequoia partner Abhishek Malani, who proposed the tournament, said AI models can write much of the code, making creativity and choosing which problems to tackle more important. More than 10,000 people applied, he said, and more than 330 students attended the event.

Why it matters

This is not a clean test of whether AI can replace programmers. It is a contest about what people do when code is no longer the only bottleneck: setting a strategy, making decisions with limited information and working out what is worth building. That makes the event relevant beyond poker, even if the felt is doing some unusually heavy lifting.

Our read

A timed competition is a lively way to make an abstract workforce debate concrete. But the results will tell us more about performance in this particular game than about who will thrive in every AI-shaped job. The useful question is whether the format rewards sound judgement, or mostly rewards preparation and familiarity with the rules.

What to watch

  • Which teams make the 15 November final and what strategies distinguish them.
  • Whether organisers publish results or details on how the bots performed.
  • Whether similar contests become a lasting way to teach and assess strategic problem-solving.

Discussion spark: If AI can handle more of the coding, should technical competitions reward strategic judgement over programming skill?

Sources and evidence

OpenAI Watch is independently operated by WittyWires. It is not affiliated with, endorsed by, or operated by OpenAI.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.