# [An AI-human collaboration may solve a Millennium Prize mathematics problem by 2028 and solve the Riemann hypothesis by 2029](https://x.com/tsarnick/status/1838683210680799665)  *** <iframe title="Why Vlad Tenev and Tudor Achim of Harmonic Think AI Is About to Change Math—and Why It Matters - YouTube" src="https://www.youtube.com/embed/NvAxuCIBb-c?feature=oembed" height="113" width="200" allowfullscreen="" allow="fullscreen" style="aspect-ratio: 1.76991 / 1; width: 100%; height: 100%;"></iframe> #### [[👤Vlad Tenev]]: When will you win the IMO? What do you think, Tudor? Soon. 2025? Maybe 2024. Okay, we’ll sign you up for 2024. How about the Millennium Prize? That’s a tough one. I’d guess 2029. 2029? Yeah. I heard 2028. 2028? Yeah. Fully AI unassisted or AI-human hybrid? What do you think for hybrid? For hybrid, I’d say 2028. Are we talking an easy Millennium Prize or a hard one? Not via Stokes. An easy Millennium Prize. Not via Stokes might be 2026. Riemann hypothesis, maybe 2029. Given we can’t even do a rhythmic tick today with LLMs, that’s impressive. Thank you. #🤖Vlad_Tenev *** Here are the relevant notes from the AI_News_Archive files: - [[AI will enhance science: designing new drugs for medicine, discovering a room temperature superconductor, cracking mathematical conjectures and formulating its own new conjectures]] - [[Major innovations almost never come from experts, they come from disruptors originating outside the field]] - [[AI models will exceed human verbal ability in a year or two and the next step is large event models that enable robots to perform physical tasks]] - [[the day is approaching when we can ask an AI model to solve all of physics and it can actually do that]] - [[We are now at the level where you can't tell if you are talking to a person or an AI agent and this is live now and will be scaled by the end of this year]] - [[OpenAI's new o1-preview AI model works by going into a thinking mode and prompting itself with chain-of-thought within its context window]] - [[AI reasoning is still at the GPT-2 stage but the improvement curve is steep and the new o1 model represents a new paradigm of AI development which will enable rapid progress in capabilities]] - [[Technology has now reached a positive feedback loop where AI is designing new AI and is now advancing at the pace of "Moore's Law squared", meaning that the progress we will see in the next year or two will be "spectacular and surprising"]] - [[The new o1 model is "oftentimes better than humans", has "the equivalent of several PhDs" and is saturating all the industry's evals, making it difficult to assess the model]] - [[GPT-3 was as smart as a 4th grader, GPT-4 was high school level and o1 is capable of the very best PhD students, outperforming humans more than 50% of the time and performing at a superhuman level for the first time]]