Max Jaderberg & Tuomas Sandholm | On AlphaFold 3, game theory application, and the future of science

This fireside discussion between Max Jaderberg and Tuomas Sandholm took place at Sana AI Summit on May 15 2024 in Stockholm, Sweden. The conversation was moderated by Sana founder and CEO Joel Hellermark.
Max Jaderberg is Chief AI Officer at Isomorphic Labs, where he leads the research and application of machine learning for drug discovery, including the latest AlphaFold model. Prio to Iso, Max pioneered numerous algorithms at DeepMind by combining large-scale deep learning, reinforcement learning, and generative models. He was also the CEO and co-founder of the image recognition company Vision Factory.
Tuomas Sandholm is a prolific professor and entrepreneur. He’s published over 500 peer-reviewed papers, holds 29 patents, and applied AI to everything from $60 billion of the world’s most complex combinatorial auctions to running the US National Kidney Exchange. He also knows a thing or two about games-having led a research group for over 20 years that developed the first superhuman AIs for two-player and multi-player no-limit Texas hold’em. Tuomas is on a mission to prove that game theory is directly useful by applying computational game theory to many high-stakes defense settings.
Joel Hellermark is the founder and CEO of Sana. An enterprising child, Joel taught himself to code in C at age 13 and founded his first company, a video recommendation technology, at 16. In 2021, Joel topped the Forbes 30 Under 30. This year, Sana was recognized on the Forbes AI 50 as one of the startups developing the most promising business use cases of artificial intelligence.
Timestamps
MC speaker introduction (00:00:00)
Starcraft to medicine (00:03:51)
AlphaFold’s breakthrough (00:06:11)
Futuristic ideas today (00:09:16)
ChatGPT’s shortcomings (00:10:57)
AI beating humans (00:12:58)
Poker tactics (00:15:14)
Real-world applicability (00:16:41)
ChatGPT can’t play poker (00:17:57)
Augmenting LLMs (00:19:00)
Move 37 (00:21:15)
AI’s disruption (00:24:50)
Constraints of AI (00:25:55)
Road to AGI (00:27:30)
Year 2034 (00:29:38)
Follow Sana
X - x.com/sanalabs
LinkedIn - / sana-labs
Instagram - / sanalabs
Try Sana AI for free - sana.ai

[@Terrahpin]

28:05 my guy is smirking like deepmind is already sitting on AGI without any of these spwcialized models he is talking about 😂

[@ODSD_EXCITEMENT]

The rest of us have to hold onto what's actually real. Trying using GPT and gemini OOTB for something serious like, help me with symbolic Reasoning given these parameters and you run hard into the grounded bits of truth, axiomatic reasoning, substantive notions of truth and fuzzy, second order logics. You get failure.