Tetragrammaton with Rick Rubin

Guest: Greg Brockman (Part 2)
Date: February 28, 2026

Episode Overview

This episode features a deep and wide-ranging conversation between Rick Rubin and Greg Brockman, co-founder of OpenAI. The discussion follows Greg's personal journey from his early interest in technology and artificial intelligence (AI) to the founding and growth of OpenAI. The episode dives into key breakthroughs in AI, the evolution of machine learning models, internal dynamics at OpenAI, and the societal implications of rapid advances in artificial intelligence.

Central themes:

• Personal motivation and formative experiences in building and coding
• Historic turning points in AI (notably deep learning and neural nets)
• OpenAI’s ethos, founding, and internal team chemistry
• The changing landscape of programming and work via AI agents
• Philosophical questions around the nature of intelligence, math, and agency
• Practical insights into how AI models are developed, trained, and evaluated
• Reflections on the industry competition and the future of knowledge work

Key Discussion Points

1. Greg’s Early Coding & "Reverse Turing Test" Project

• [00:23-02:46] Greg's first impactful programming project was a game based on the "reverse Turing test," inspired by Alan Turing’s 1950 paper.
  • He created an online game where two humans compete to identify each other while also conversing with an AI.
  • Quote:

    “I taught myself how to code... And I remember that I built this game... a two player game. So I was sitting in the lobby... just sadly waiting for someone to show up... One day... I got 1500 hits from StumbleUpon. It was an amazing moment.” — Greg Brockman [01:12]

  • Experience of "something in my head is now in reality," chased by many builders.

2. Turing’s Prediction & AI’s Evolution

• [02:49-04:20] The foundational influence of Turing’s vision, especially the need for machines that can "learn" rather than follow set rules.
  • Quote:

    “If you can build a machine... that learns like a human child... and you can then have a human who gives it rewards and punishments... then that's how you will solve this test. And here's the wild thing. That is exactly what we've been doing.” — Greg Brockman [03:26]

• [04:23-05:49] The key constraint on AI progress: computational power, not conceptual limitations.

3. History of Deep Learning & the AlexNet Revolution

• [08:15-13:51] Explains the significance of the AlexNet breakthrough in 2012, which “cemented the deep learning revolution.”
  • Neural nets went from fringe to the field’s focus due to their overwhelming success in the ImageNet competition.
  • Quote:

    “They won this competition... it just blew everything else out of the water, like a massive jump.” — Greg Brockman [11:09]

• The academia vs. fringe struggle: neural nets repeatedly dismissed until democratization of compute allowed them to resurface.
  • Quote:

    “There was a very concerted campaign waged against this whole direction... the symbolic systems people got in very cozy with funding agencies... and that’s what killed it in the '70s.” — Greg Brockman [15:09]

4. OpenAI’s Early Days and Team Dynamics

• [17:13-18:27] OpenAI’s beginnings—literally in Greg's San Francisco living room.

  • Describes the day-one team, "a big open space and we had a black wood table... some couches... a big screen TV…”

• The founding team’s ethos: build AGI as a positive force for humanity, even before figuring out 'how.'

• [21:41-27:42] The significance of researcher-engineer relationships in AI’s progress at OpenAI; importance of tight, iterative collaboration over rigid division of labor.

  • Quote:

    “What I usually do when working with [Wojciech] is really think about, okay, what are the bounds on what we should think about?” — Greg Brockman [28:24]

• [25:23-27:42] Recounts the offsite in Napa that unified the founding team and set their three-part plan:

  1. Solve reinforcement learning (RL)
  2. Solve unsupervised learning (UL)
  3. Gradually tackle more complexity

5. From Stripe to OpenAI: Personal Path and Tech Philosophy

• [28:42-40:42] Greg discusses his journey from MIT to Stripe, driven by both pattern-matching from peers and a desire to solve "real problems" despite not being passionate about payments.
  • Quote:

    “I had gotten this company to a place where it was going to succeed with or without me. And then the question was, do I continue or do I go and do something new?” — Greg Brockman [39:05]

• A deep reflection on carving a personal identity as an outsider and the power of working with peers who "click" on a technical and personal level.

6. Coding as Art, Math, and Magic

• [46:09-51:12] Coding compared to mathematics and management; the magic of translating a mental vision into reality.
  • Quote:

    “It's almost like magic. You sort of have this vision in your head, and you just, by describing it, somehow it comes to be.” — Greg Brockman [47:13]

• The philosophical debate: Is math a fabric of the universe, or an overlay from observers? Are we discovering or inventing mathematics?

7. Modern Programming & "Vibe Coding" with AI

• [53:36-55:52] "Vibe coding" defined—programming by instructing AI in natural language rather than writing code directly.

  • Recent improvements in AI models have made powerful “agentic” programming available to expert engineers.
  • Quote:

    “Vibe coding is... moving the machine closer to the human... you're acting as a manager who's still very accountable for the outcome.” — Greg Brockman [54:03]

• [57:29-59:46] The rising “sea level" of model capabilities—what used to take months now takes minutes.

  • The future: humans as “managers of agents,” focusing on outcomes rather than mechanical details.

8. The Future of Agents, Code Quality, and Human Oversight

• [60:10-64:27] OpenAI’s "agent-first" approach: shifting from text editors and terminals to intelligent digital agents.
  • Emphasis on building trust and quality—"say no to slop" in code, holding AIs to an even higher standard than humans in code reviews.
  • The critical metric for future productivity: “how much compute does an individual human marshal.”

9. Technological Revolutions and Anticipating the Future

• [64:34-68:30] Reflection on the major technological shifts, FOMO, and learning that the world repeatedly produces “magical” innovations that seem impossible to previous generations.

• [69:34-71:03] AGI redefined: not just as human-level task ability, but as an “individual force multiplier,” empowering a person to act as "the visionary, the CEO."

10. OpenAI’s Product Suite and Agentic AI

• [71:08-73:48] ChatGPT as one expression of OpenAI's goal—"intelligence on demand for your problem."

  • Codex as the general-purpose agent harness, applied beyond coding to any knowledge work.

• Definition of “agentic AI”: Models that take actions via tools, not just generate text.

  • Quote:

    “It’s actually able to take action... For agentic AI that its actually embedded in the real world and can take action.” — Greg Brockman [74:00]

11. How Large Language Models Actually Work: Pre-training & Post-training

• [73:48-83:53] In-depth explanation on how large models are pre-trained on massive datasets and then post-trained to steer behavior.
  • Pre-training is like “the Library of Congress”; post-training prunes, targets, or specializes.
  • The shift from manual, large-scale data labeling to targeting the highest-value domain expert feedback as models improve.
  • Training now includes reinforcement learning, allowing models to discover new knowledge via experimentation—unlocking AlphaGo-style moments in other domains.

12. AI Safety vs. Breakthroughs: Can AI Challenge the Status Quo?

• [85:21-89:18] The balance between steering AI for safety and enabling it to discover radically new knowledge.
  • Quote:

    “The bigger framework... really is that this technology, we have the ability to steer it... There’s a lot of antibodies that try to prevent that. But that is also how society moves forward." — Greg Brockman [89:18]

• Example: AI disproving an accepted physics hypothesis, published with a previously skeptical professor.

13. Compute as a Human Right & The Coming Compute Crunch

• [91:01-93:56] The costs and bottlenecks of massive AI models: compute, electricity, and hardware.

  • Prediction: “Compute will be a basic human right... The more compute they have, the higher quality of life they can have.”

• Discusses hardware evolution—GPUs may be replaced as the AI “manufacturing pipeline” gets more efficient.

14. Industry Competition and OpenAI’s Differentiation

• [94:31-96:58] Perspectives on Anthropic, Google Gemini, Grok, etc.; OpenAI’s focus is long-horizon, basic research and paradigm shifts, not feature chasing.
  • Quote:

    “We invest the most in basic research, in actual paradigm shifts, and you can see this...” — Greg Brockman [95:01]

15. Benchmarks: Proxy vs. Reality in AI Progress

• [97:08-99:13] On the limitations of optimizing models for academic benchmarks; shifting focus to real-world utility and usage as the "ultimate benchmark."

16. Developer Roles and Research-Deployment Continuum

• [99:20-102:12] Developers at OpenAI move between research (creating new models) and deployment (bringing them to users); successful deployment demands deep mutual understanding.

17. Open Claw/Cloudbot and "Hacker Spirit"

• [102:12-105:25] Greg praises the hacker ethos of projects like Open Claw, while noting the urgent need for more robust trust and safety architectures as agents move closer to real-world action.

18. OpenAI’s Philosophy: Technology by and for Everyone

• The importance of public transparency and inclusion in AI development:

  “If you’re going to build technology that’s going to change everyone’s lives, I think people need to know about it, right? People need to be included in that.” — Greg Brockman [104:24]

Memorable Quotes

• On Building the Future:

  “That feeling that this thing was in my head and now it’s in reality—and now these people are all enjoying what I built—and I want to keep chasing it.” — Greg Brockman [01:45]

• On Turing’s Foresight:

  “He is just so smart. And that's why. Yeah, he. He.” — Greg Brockman reflecting on Alan Turing [04:12]

• On Neural Nets’ Rebirth:

  “The neural net people knew what they wanted to do, they wanted bigger computers, deeper neural nets.” — Greg Brockman [15:09]

• On OpenAI’s Early Team:

  “It was a big open space and we had a black wood table that was this big oval shape... Day one, there was no whiteboard.” — Greg Brockman [17:22]

• On the Coding Process:

  “It's almost like magic. You sort of have this vision in your head and you just, by describing it, somehow it comes to be.” — Greg Brockman [47:13]

• On Compute as a Human Right:

  “Compute will be a basic human right... The more compute they have, the higher quality of life they can have.” — Greg Brockman [92:24]

• On OpenAI’s Motivation:

  “AGI... is something that everyone deserves to participate in.” — Greg Brockman [22:55]

Notable Segments by Timestamp

• Greg’s first viral coding experience & 'reverse Turing test': [00:23–02:46]
• Turing’s learning machine and early AI influences: [02:49–04:20]
• The AlexNet deep learning revolution: [08:15–13:51]
• OpenAI’s living room origins and first team: [17:13–18:27]
• OpenAI’s team chemistry and off-site vision: [25:23–27:42]
• Transition from Stripe to OpenAI: [28:42–40:42]
• Coding as math, management, and magic: [46:09–51:12]
• Vibe coding and feedback loops with AI: [53:36–55:52]
• Shifting human role: from coder to manager of AI agents: [57:29–59:46]
• Future of agentic software and “agent-first” organization: [60:10–64:27]
• How LLMs are trained: pre-training, post-training, RL: [73:48–83:53]
• AI outstripping textbooks (physics breakthrough): [87:37–89:18]
• Compute as society’s critical new resource: [91:01–92:46]
• Open Claw, trust, and the ethics of open deployment: [102:12–105:25]

Episode Character & Tone

• The conversation is relaxed, candid, and reflective—balancing technical depth with open-minded speculation.
• Greg’s tone is enthusiastic but grounded, generous with credit to collaborators and historical giants.
• Rick Rubin’s gentle, exploratory questioning draws out both big-picture philosophies and nitty-gritty engineering tales.

Takeaway

This episode charts the intellectual, technical, and human journey behind OpenAI and the ongoing AI revolution. Brockman brings humility, historical depth, and a sense of real urgency to the tasks ahead—with a central thesis: as AI gets more powerful, the questions shift from "can we build it?" to "will we use it wisely and for everyone’s benefit?"

For listeners or readers: this episode is an essential trip through where AI came from, how it works, and where it may soon take us—as told from inside the eye of the storm, by one ambitious builder seeking not just progress, but wisdom and inclusivity along the way.