AMA: career advice given AGI, how I research ft. Sholto & Trenton

Dwarkesh Patel, Trenton Bricken, and Sholto Douglas discuss AGI career strategy, why LLMs can’t make cross-domain discoveries, and how YouTube Shorts drove half the podcast’s growth.

• Trenton canceled his 401k contributions due to fast AGI timelines; neither Sholto nor Dwarkesh made other major lifestyle changes.
• Sholto: LLMs fail at novel cross-domain discovery because pre-training doesn’t reward it; scaled RL with real-world interaction is the missing ingredient.
• YouTube Shorts drove at least half of Dwarkesh’s podcast growth — he calls it the single most counterintuitive distribution win for long-form technical content.
• Dwarkesh’s top-performing guests are scholar Sarah Payne and geneticist David Reich, not lab CEOs — big names don’t predict viewership.
• Dwarkesh hired four editors via a public clips competition: a farmer in Argentina, a student in Sri Lanka, a former Mr. Beast editor, and a Czech AI-animation director.
• Career advice given AGI: deep technical expertise still matters because individuals will soon be managing large teams of AI agents, not just using them as assistants.
• Despite ~1,000 applications, Dwarkesh’s best hire came through a mutual referral — he concludes top candidates rarely apply to postings.
• Book ‘The Scaling Era’ (Stripe Press) includes two unpublished interviews, including Jared Kaplan explaining scaling through physics and data-manifold framing.

2025-03-25 · Watch on YouTube