How Game Theory Can Make AI More Reliable

The original version of this story appeared in Quanta Magazine.

Imagine you had a friend who gave different answers to the same question, depending on how you asked it. “What’s the capital of Peru?” would get one answer, and “Is Lima the capital of Peru?” would get another. You’d probably be a little worried about your friend’s mental faculties, and you’d almost certainly find it hard to trust any answer they gave.

That’s exactly what’s happening with many large language models (LLMs), the ultra-powerful machine learning tools that power ChatGPT and other marvels of artificial intelligence. A generative question, which is open-ended, yields one answer, and a discriminative question, which involves

→ Continue reading at Wired - Science

More from author

Related posts


Latest posts

The Science of Having a Great Conversation

If you’ve ever spoken to someone and later felt that you would have better spent your time talking to a brick wall, you’ll surely...

Fake beauty queens charm judges at the Miss AI pageant

Aiyana Rainbow, a Romanian-made AI model, is one of the Miss AI finalists....

Apple needs to make the iPhone cool again. Tomorrow is its chance | CNN Business

New York CNN  —  Apple hasn’t given users a significant reason to buy an upgraded iPhone for four years,...