Steve Hsu - Intelligence, Embryo Selection, & The Future of Humanity

ℹ️ EPISODE INFO

Released

August 23, 2022

Duration

2h 21m

Channel

Dwarkesh Podcast

Watch on YouTube

▶ Open ↗

📝 EPISODE DESCRIPTION

Steve Hsu is a Professor of Theoretical Physics and of Computational Mathematics, Science, and Engineering at Michigan State University, and one of the founders of the company Genomic Prediction. We go deep into the weeds on how embryo selection can make babies healthier and smarter. Steve also explains the advice Richard Feynman gave him to pick up girls, the genetics of aging and intelligence, and the psychometric differences between shape rotators and wordcels. Read Transcript: https://www.dwarkeshpatel.com/p/steve-hsu Apple Podcasts: https://apple.co/3wob9AK Spotify: https://spoti.fi/3PCNK5m Steve Hsu's Blog: infoproc.blogspot.com Follow Steve: https://twitter.com/hsu_steve Follow me: https://twitter.com/dwarkesh_sp TIMESTAMPS 0:00:00 Intro 0:00:49 Feynman’s advice on picking up women 0:12:21 Embryo selection 0:24:54 Why hasn't natural selection already optimized humans? 0:34:48 Aging 0:43:53 First Mover Advantage 0:53:50 Genomics in dating 1:00:32 Ancestral populations 1:07:59 Is this eugenics? 1:16:00 Tradeoffs to intelligence 1:25:02 Consumer preferences 1:30:15 Gwern 1:34:36 Will parents matter? 1:45:26 Word cells and shape rotators 1:57:27 Bezos and brilliant physicists 2:10:24 Elite education

🎤 SPEAKERS

• Steve Hsu

  guest

• Dwarkesh Patel

  host

• Narrator

  other

🧾 EPISODE SUMMARY

In this episode of Dwarkesh Podcast, featuring Steve Hsu and Dwarkesh Patel, Steve Hsu - Intelligence, Embryo Selection, & The Future of Humanity explores genetic Prediction, IVF, and Engineering Intelligence: Steve Hsu’s Vision Steve Hsu discusses how advances in genomics, AI, and IVF enable embryo selection based on polygenic predictors for disease risk, health, and eventually traits like height and intelligence. He explains the mathematical and evolutionary foundations of polygenic scores, emphasizing that most complex traits are surprisingly additive and highly polygenic, making them predictable with enough data. Hsu outlines his company Genomic Prediction’s role in IVF clinics worldwide, where embryo genotyping already helps parents avoid embryos with extreme disease risks, and sketches how future CRISPR editing and better data could allow far more ambitious trait optimization. The conversation also ranges into ethical and societal implications—inequality, regulation, national strategies—plus digressions on Feynman, elite education, talent, and the strengths of physics-style quantitative thinking.

🔗 RELATED EPISODES

The better AI gets, the smaller its share of the economy might get – Alex Imas and Phil Trammell

What rebuilding AlphaGo teaches us about self-play, RL, and future of LLMs - Eric Jang

How GPT, Claude, and Gemini are actually trained and served – Reiner Pope

Jensen Huang – Will Nvidia’s moat persist?

Terence Tao – How the world’s top mathematician uses AI

Elon Musk – "In 36 months, the cheapest place to put AI will be space”