Episode 351

Artificial intelligence is starting to solve mathematical theorems better than humans. Mathematicians say AI is now an existential threat to their work. As one professor puts it; “We are running out of places to hide.”

From winning gold medals at mathematics competitions, to solving previously unanswered Erdős problems, multiple AI achievements have come together recently to exceed all expectations of its capabilities.

Find out just how quickly the tech is advancing, how we can tell the AI isn’t just hallucinating answers, why it may help us formalise all of mathematics - and whether it will really put humans out of a job.

And 10 years on since Google’s AlphaGo AI first beat human Go master Lee Sedol, we reflect on that epic moment and hear from Chris Maddison who saw it all unfold.

Rowan Hooper is joined by New Scientist’s Alex Wilkins to discuss “one of the most remarkable stories” he’s ever worked on.

Chapters

(00:00) Intro - The biggest moment in the history of mathematics

(01:10) The many problems AI is now solving

(04:11) Are these models similar to ChatGPT or Claude?

(05:09) Will AI help us advance the field of mathematics?

(07:28) How can we check AI’s answers - are they just hallucinations?

(10:51) Why it’s important to “formalise” maths

(12:03) Will we become too reliant on this AI?

(13:00) 10 years on since AI beat Lee Sedol at Go

(14:54) AI creativity: The famous ‘Move 37’

(16:50) How it felt to watch this epic moment

(19:21) How AlphaGo led to the LLMs of today

(20:25) Are regular chatbots becoming more creative?

To read more about these stories, visit https://www.newscientist.com/

Learn more about your ad choices. Visit megaphone.fm/adchoices