The AI Impact on Math | with Pr. Nourdin author of Math Research with GPT5

Did GPT-5 just discover new mathematics and start competing with human mathematicians? In this episode of The AI Impact Podcast, we dive into the AI impact on mathematics with Professor Ivan Nourdin – professor of mathematics at the University of Luxembourg, head of their bachelor program in mathematics, respected researcher and book author, and co-founder of the AI startup Grew AI, which helps public and private organisations use AI responsibly. A few months ago, Sebastian Bubeck (OpenAI, and a leading voice in AI) posted on X that GPT-5 Pro had found a new proof in convex optimization in just a few minutes. The post went viral, with many people claiming that AI can now do “real research-level math”, maybe even replacing mathematicians. Ivan didn’t just react on social media – he systematically tested these claims: he tried to replicate Bubeck’s experiment, and then asked GPT-5 a different question: to extend a result in a mathematical theory he co-created 20 years ago, to see whether it could really produce new proofs. In this conversation, we explore: What actually happened in Bubeck’s GPT-5 experiment – and why it matters Ivan’s own research with GPT-5 on advanced mathematics: where it impressed him, and where it failed Why almost correct is not enough in math (and how a single tiny mistake collapses a proof) How generative AI can realistically help mathematicians today – and where the hype goes too far -The impact on PhD training, students, and education: can you still learn to think deeply if you shortcut everything through AI? -The role of proof assistants like Lean, and why verification is becoming a huge challenge as mathematical output explodes Whether we should be hopeful or worried about a future where AI does more and more of the work in mathematics If you care about discovery, creativity, and the future of human expertise in the age of AI, this episode is for you.