No Priors Podcast: "The Best of 2025 (So Far)"

Hosts: Sarah Guo & Elad Gil
Date: October 31, 2025
Theme:
A curated "Best Of" episode featuring transformative moments and perspectives from top AI leaders, founders, and researchers. The episode explores breakthroughs in AI capability, the impact on work and society, paradigm shifts in reasoning, and deeply human stories arising from the AI revolution.

Key Discussion Points & Insights

1. Accidental Beginnings and Early Legal AI (Harvey: Winston Weinberg)

• Timestamp: 00:30 – 02:02
• Insight: Sometimes, recognizing overlooked capabilities is the spark for disruptive startups.
  • Winston describes stumbling into AI for legal workflows, noticing "no one was talking about GPT-3," and being surprised by its accuracy in legal advice.
• Experiment: They took 100 landlord-tenant questions from Reddit, generated AI-powered answers using chain-of-thought prompts (before such techniques were widely known), and let attorneys review the responses.
  • 86% were deemed good enough to be sent to clients unedited.
• Notable Quote:
  • "We cold emailed the General Counsel of OpenAI... his response basically was, 'Oh, I had no idea the models were this good at legal.'" (01:28)
• Impact: This early validation led directly to high-level engagement with OpenAI.

2. The Challenge of Spatial Intelligence (Dr. Fei-Fei Li, AI Luminary)

• Timestamp: 02:10 – 04:12
• Insight: Spatial intelligence—the ability to model and interact with 3D environments—is primitive even in humans and animals, making it a major frontier for AI.
• Memorable Analogy:
  • "If I ask you to close your eyes right now and draw out or build a 3D model of the environment around you, it's not that easy... 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. No pun intended." (03:25)
• Implication: Mastering this will unlock new and transformative forms of interaction.

3. AI, Automation, and the Evolving Workforce (Brendan Foody, Merck War CEO)

• Timestamp: 04:24 – 06:10
• Prediction: Rapid, painful displacement of many jobs is coming, especially in digital roles, leading to political and societal upheaval.
  • The economy must figure out new purposes for customer support, recruiting, and other soon-to-be-automated workers.
  • Automation in physical jobs (robotics, therapy, personal interaction) will lag digital automation.
• Notable Quote:
  • "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... it's going to happen." (04:24)
  • "Automation in the physical world is going to happen a lot slower than what's happening in the digital world." (05:53)

4. Geopolitics and the Risks of Superintelligence (Dan Hendricks, Center for AI Safety)

• Timestamp: 06:20 – 08:05
• Insight: As superintelligence becomes feasible, global stability could hinge on AI. Countries may be compelled to preemptive cyberattacks to avoid being overtaken.
• Comparison: Nuclear deterrence as precedent, but AI development could be even more destabilizing and opaque.
• Notable Quote:
  • "Later on it becomes so destabilizing that China just says we're going to do something preemptive like do a cyber attack on your data center. And the US might do that to China..." (07:20)

5. Making Entrepreneurship Scientific (Noubar Afeyan, Flagship Pioneering/Moderna)

• Timestamp: 08:21 – 10:37
• Philosophy: Challenges the norm that entrepreneurship should be random and "gamey." Wants to transform it into a scientific, professional discipline—especially in biotech and high-stakes industries.
• Notable Quote:
  • "Entrepreneurship... was supposed to be random, improvisational, kind of idiosyncratic, almost emotional, gamey. All of those things I kind of thought was a bit of a put off when it comes to actually doing things in a serious professional way." (08:54)
  • "You can't think of this as like shots on goal... we can get better." (10:34)

6. Paradigm Shift: Reasoning and Tool Use in AI (OpenAI: Brandon McKinsey, Eric Mitchell; Harvey: Winston Weinberg)

• Timestamp: 10:46 – 12:41
• Insight: New models can recognize their own uncertainty; tool use (like code, image manipulation) lets them overcome limitations and reason more effectively.
• Notable Quote (McKinsey):
  • "The model will very transparently tell you initiative thought, like, I don't know, I can't really see the thing you're talking about very well... when you give it access to a tool... it's much more productive use of tokens." (11:00)
• Notable Quote (Weinberg):
  • "You can just allocate compute a lot more efficiently because you can defer stuff that the model doesn't have comparative advantage to doing to a tool that is really well suited to doing that thing." (12:30)

7. The Grind: Taste, Data Generation, and Hands-on Breakthroughs (OpenAI: Issa Fulford, Dan Hendricks)

• Timestamp: 12:58 – 13:49
• Insight: Progress often comes from pragmatic experimentation and sheer effort, not grand theory.
• Memorable Exchange:
  • Fulford: "It really was one of those things where we thought that training on browsing tasks would work... But actually the first time you train a model on a new data set... and seeing it actually working... was pretty incredible." (13:09)
  • Hendricks: "Sometimes it will make a mistake where it will do such smart things and then make a mistake where I'm just thinking, why are you doing that? Stop." (13:40)

8. Turning a 'Bad Market' into a Great One: Enterprise Search (Glean: Arvind Jain)

• Timestamp: 14:05 – 16:21
• Story: Glean began by recognizing that enterprise search—a "graveyard market"—was newly possible thanks to SaaS.
• Key Unlock: With SaaS, enterprise data is accessible via APIs, enabling unified search for the first time.
• Notable Quote:
  • "The enterprise world has changed... SaaS systems don't have versions. All customers have the same version. They're open, they're interoperable. You can actually hit them with APIs and get all the content. I felt that the biggest problem was actually solved." (14:40)
• Fun Fact:
  • Glean's largest customers have over 1 billion documents—comparable to the size of the early internet (2004).

9. AI's Human Impact: Doctors, Fulfillment, and Purpose (Abridge: Shiv Rao)

• Timestamp: 16:30 – 18:21
• Moving Story: Dr. Rao shares a message from a rural doctor who now gets to have dinner with her child because of AI tools automating note-taking.
• Notable Quote:
  • "My son asked me, mommy, why aren't you working right now? I literally took my phone out and explained to him that Abridge is a new tool that lets mommy come home early and eat dinner with her family..." (17:15)
  • "We get feedback like that, like every day... it's the oxytocin hits like this. It's the purpose, it's the fulfillment... that's what we're really after." (17:52)

Notable Quotes (with Speaker & Timestamp)

• Winston Weinberg (Harvey):
  "We cold emailed the General Counsel of OpenAI... his response basically was, 'Oh, I had no idea the models were this good at legal.'" (01:28)
• Dr. Fei-Fei Li:
  "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. No pun intended." (03:36)
• Brendan Foody (Merck War):
  "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..." (04:24)
• Dan Hendricks (AI Safety):
  "Later on, it becomes so destabilizing that China just says we're going to do something preemptive like do a cyber attack on your data center. And the US might do that to China..." (07:20)
• Noubar Afeyan (Flagship Pioneering/Moderna):
  "You can't think of this as like shots on goal... we can get better." (10:34)
• Brandon McKinsey (OpenAI):
  "The model will very transparently tell you... I don't know, I can't really see the thing you're talking about very well. Or like, it almost knows... that its vision is not, not very good." (11:00)
• Winston Weinberg (Harvey):
  "You can just allocate compute a lot more efficiently because you can defer stuff that the model doesn't have comparative advantage to doing to a tool..." (12:30)
• Arvind Jain (Glean):
  "SaaS systems don't have versions. All customers have the same version. They're open, interoperable. You can actually hit them with APIs and get all the content. I felt that the biggest problem was actually solved." (14:40)
• Shiv Rao (Abridge):
  "...Abridge is a new tool that lets mommy come home early and eat dinner with her family..." (17:15) "It's the purpose, it's the fulfillment... that's what we're really after." (17:52)

Memorable Moments & Segment Highlights

• The serendipitous "aha" behind Harvey’s founding (00:30)
• Fei-Fei Li on the evolutionary roots and limits of spatial intelligence (02:10)
• Political risks and the possibility of AI-led cyber conflict (06:20)
• Entrepreneurship as a science, not a lottery (08:21)
• Real-world impact: Doctor’s child grateful for AI (16:30)

Takeaway

The "Best of 2025 (So Far)" episode of No Priors is a sweeping tour of how AI is quietly and quickly reshaping industries, work, global stability, and even family life.
From technical paradigm shifts and new entrepreneurial philosophies to intimate moments of joy and relief enabled by AI, the episode underscores the inflection point we collectively inhabit—and the urgency and promise that come with it.