No Priors Ep. 138 | The Best of 2025 (So Far) with Sarah Guo and Elad Gil

2025 has thus far been a year of great leaps and advances in AI technology. And Sarah and Elad have spoken with some of the most enterprising founders and scientific minds in the field of AI today. So we’re revisiting a few of our favorite conversations on No Priors so far in 2025 – Winston Weinberg (Harvey), Dr. Fei-Fei Li (World Labs), Brendan Foody (Mercor), Dan Hendrycks (Center for AI Safety), Noubar Afeyan (Flagship Pioneering), Brandon McKinzie and Eric Mitchell (OpenAI o3), Isa Fulford (OpenAI), Arvind Jain (Glen), and Dr. Shiv Rao (Abridge). Sign up for new podcasts every week. Email feedback to show@no-priors.com Follow us on Twitter: @NoPriorsPod | @Saranormous | @EladGil Chapters: 00:00 – Episode Introduction 0:21 – Winston Weinberg on Leaning into New Capabilities 02:01 – Dr. Fei-Fei Li on Spatial Intelligence 04:13 – Brendan Foody on AI Disruption in the Workforce 06:10 – Dan Hendrycks on the Geopolitics of Superintelligence 08:06 – Noubar Afeyan on Entrepreneurship 10:38 – Brandon McKinzie and Eric Mitchell on Reasoning Models 12:41 – Isa Fulford on Training Deep Research 13:49 – Arvind Jain on Innovating Enterprise Search 16:21 – Dr. Shiv Rao on AI’s Human Impact 18:58 – Conclusion

Sarah GuohostWinston WeinbergguestFei-Fei LiguestBrendan FoodyguestElad GilhostDan HendrycksguestNoubar AfeyanguestBrandon McKinzieguestEric MitchellguestIsa FulfordguestGuest (founder describing building a search product)guest

Oct 30, 202518mWatch on YouTube ↗

WHAT IT’S REALLY ABOUT

AI’s Hinge Moment: Work, Wealth, Safety, and Human Impact in 2025

This highlight episode of No Priors showcases pivotal 2025 conversations about how AI is reshaping industries, work, and global dynamics. Founders and researchers from Harvey, OpenAI, Glean, Abridge, and others describe discovering hidden opportunities as models quietly became good enough to transform “bad” markets like legal advice and enterprise search. Guests explore frontier capabilities such as spatial and visual reasoning, tool-using models, and AI-assisted research, alongside the labor displacement and geopolitical risks of approaching superintelligence. The episode closes on the deeply human side of AI, with healthcare AI enabling doctors to reclaim time, purpose, and connection with their families.

IDEAS WORTH REMEMBERING

5 ideas

Hidden AI capabilities can unlock ‘impossible’ markets when paired with the right workflow.

Harvey’s early GPT-3 experiments in legal Q&A showed attorneys would ethically send most AI-generated answers unchanged, revealing a large, underexploited opportunity in legal reasoning before the wider market recognized it.

Spatial and visual reasoning remain hard problems where AI could surpass everyday human abilities.

Fei-Fei Li notes that even trained humans struggle to mentally model 3D environments, suggesting huge potential for AI tools that make 3D understanding, manipulation, and editing easy and fluid for non-experts.

AI will rapidly displace many digital roles, pushing more human work into the physical and interpersonal world.

Brendan Foody anticipates quick, painful displacement in fields like customer support and recruiting, with slower automation in physical and human-interaction jobs such as robotics data collection, hospitality, and therapy.

Superintelligence raises destabilizing geopolitical dynamics reminiscent of nuclear deterrence.

Dan Hendricks argues that as AI becomes pivotal to national power, states may both deter each other from AI ‘first strikes’ and consider preemptive cyber operations on rivals’ data centers to avoid being technologically crushed.

Entrepreneurship in high-stakes areas like health and climate can be made more systematic and less random, with AI as a key enabler.

Noubar Afeyan critiques the ‘gamey’ culture of startups and advocates for treating company creation as a professional, methodical discipline, where AI can help design and validate ambitious biotech and deep-tech concepts more reliably.

WORDS WORTH SAVING

5 quotes

We took about 100 landlord-tenant questions… and 86 out of 100 was yes.

— Winston Weinberg (Harvey CEO)

Imagine you do it at your fingertip much more easily and allow much more fluid interactivity and editability. That would just be a whole different world for people.

— Dr. Fei-Fei Li

I think displacement in a lot of roles is going to happen very quickly, and it's going to be very painful, and a large political problem.

— Brendan Foody (Merck Core CEO)

Later on, it becomes so destabilizing that China just says, 'We're going to do something preemptive, like do a cyberattack on your data center,' and the US might do that to China.

— Dan Hendricks (Center for AI Safety Director)

Mommy's gonna be able to eat dinner with us every night now.

— Doctor at Tanner Health, via Shiv Rao (Abridge CEO)

Emergence of practical AI capabilities in domains like law and enterprise searchSpatial and visual reasoning as a frontier of AI intelligenceLabor displacement, new kinds of work, and wealth reallocation in an AI economyGeopolitical and safety concerns around superintelligence and AI arms racesSystematizing entrepreneurship and biotech innovation with AIReasoning models, tool use, and test-time scaling in modern AI systemsHuman-centered AI in healthcare and its impact on clinician wellbeing

High quality AI-generated summary created from speaker-labeled transcript.

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.