Shams was trained in theoretical physics before becoming an AI engineer at DeepMind and founding the company Mutable, which was acquired by Google. He is now lead engineer for Google's software agent group.
Steve and Omar discuss:
(00:00) - Introduction
(01:18) - Journey from Physics to AI
(10:51) - Elon Tried to Buy DeepMind
(16:52) - Building Mutable and Auto Wiki as Context for LLMs
(33:39) - The Value of AI Talent and Meta's AI Hiring Spree
(42:58) - AI and The Workforce
(58:02) - The Intersection of Physics and AI
Audio-only version and transcript:
https://www.manifold1.com/episodes/omar-shams-ai-founder-and-google-ai-agent-lead-89
I definitely agree that training in physics is remarkably useful for making progress in other areas, like AI or biology, which came up in this interview and others. I doubt anyone disagrees, and a lot of these other areas are growing, yet the number of physics majors has been declining. (In the US, at least.) "For the third year in a row, the number of physics bachelor’s degrees awarded by US institutions has declined." [https://pubs.aip.org/physicstoday/article-abstract/77/12/26/3321124/Physics-bachelor-s-degrees-in-the-US-trend] In fairness, Physics rose in popularity from about 2010-2020. Still, the current trend is disturbing, and many physics departments are struggling with low numbers of majors. Here (your former home, U. of Oregon) we awarded 21 undergraduate degrees this year; the total student population is about 20,000. There are a lot of reasons that physics isn't as popular as it should be, spanning high school curricula, undergraduate advising, math education, and more, but that's another story.
Thanks for the interview; interesting as always!