Big Technology Podcast — OpenAI President Greg Brockman: AI Self-Improvement, The Superapp Bet, Path to AGI, Scaling Compute

Host: Alex Kantrowitz
Guest: Greg Brockman, Co-founder and President of OpenAI
Date: April 1, 2026

Executive Overview

In this insight-packed episode, Alex Kantrowitz sits down with Greg Brockman, president and co-founder of OpenAI, for an in-depth conversation at a pivotal moment in the company’s history. The discussion explores OpenAI’s strategic shift towards prioritizing foundational applications, the launch of a “super app,” challenges and advances in scaling AI and compute, the journey to AGI (artificial general intelligence), and AI’s societal impact.

Brockman candidly shares stories from OpenAI’s meteoric rise, offers a behind-the-scenes look at pivotal decisions, and weighs in on AGI's imminent arrival. The conversation ranges from technical trajectories and infrastructure bets to the ethics of AI development and the politics of technological transformation.

Key Discussion Points & Insights

OpenAI’s Strategic Shift: From Video to the Super App

[01:02-04:06]

• OpenAI is shutting down video generation (Sora) and channeling resources into developing a “super app” focused on business and coding (not strictly moving from consumer to B2B, but prioritizing applications with highest potential impact).
• The pivot reflects both advances in deep learning and practical constraints on compute: “No matter how much we try to build, I know we’re not going to be able to keep up with the demand.” (Brockman, [00:00])
• Two applications are “at the top of the stack rank”: a personal assistant (aligned, trustworthy, knows you) and an AI that can solve hard problems for you across work/life contexts.

Technology Choices: The Model Tree & Reasoning Bets

[04:44–11:19]

• OpenAI’s main bet is on reasoning models (the GPT series) rather than “world models” (like Sora for video).
• Video and robotics are transformative but require a separate tech branch (and more compute than is currently feasible for deployment at scale): “Pursuing both branches is very hard for us to do for these applications.” (Brockman, [04:44])
• The company is focusing on unified models, enhancing capabilities like bidirectional communication, speech, and multimodality, all within the GPT lineage.

The Super App: Vision, Use Cases, and Rollout

[11:24–16:52]

• The upcoming super app will bring together ChatGPT, Codex (for coding and automation), and browser integration, serving both personal and business users.
• It seeks to be the "endpoint application" for experiencing AGI: “Anything you want your computer to do, you can ask it.” (Brockman, [13:35])
• Key features: persistent memory, contextual knowledge (e.g., calendar, email), and unified agentic workflows.
• Rollout will be incremental over “the next couple months,” starting with an enhanced Codex app for general knowledge work.

Competition and Company Culture

[18:35–22:59]

• Anthropic’s Claude suite prompted OpenAI to refocus on last-mile usability and bridging smart models with real-world, messy codebases.
• Brockman dismisses any sense of OpenAI resting on its laurels: “We never feel complacent… For me, it’s been a very healthy thing to see competitors emerge and do a good job.” ([21:15])
• The company is now integrating research and deployment tightly, fostering alignment and focus.

Model Advances: The SPUD Model and What’s Next

[22:59–28:14]

• SPUD is the latest base model — “two years’ worth of research coming to fruition.”
• Expect substantial leaps in capability, nuance, and contextual understanding.
• Brockman notes the user-perceived improvement may be gradual but will become more evident as mental models update: “What we’re going to see is that for any application like that, it’s going to become so much more evident to everyone that the AI can help you.” ([26:43])

AI as Automated Researcher & Accelerating Takeoff

[28:14–31:39]

• OpenAI is building an automated AI researcher to accelerate internal progress; the tool will take on end-to-end research scientist workflows under human direction: “Take the full end-to-end of what one of our research scientists does and be able to do that in silicon.” ([31:01])
• The concept of “takeoff”: growing AI capability feeding back into itself, leading to rapid societal and economic impact.

Safety, Race Dynamics, and Societal Infrastructure

[31:39–36:00]

• Brockman stresses investment in safety, especially around tool use and prompt injections: “We invest a lot in safety, security. Good example is prompt injections…” ([31:39])
• He frames the “race” for AI broadly, emphasizing the need for societal resilience: “If this technology is going to come and change everything for everyone, people need to participate in that. It can’t be something that’s done off in secret by just one sort of centralized group.” ([35:00])

AGI Timeline & Definitions

[36:00–38:00]

• Brockman believes AGI is within a couple years: “We are going to have AGI within the next couple years in a way that is still going to be jagged.” ([00:00], [37:18])
• AGI capabilities are “jagged” — superhuman at some tasks, subhuman at others; progress is more a “vibe” than a hard science threshold.

Autonomy, Agents, and the Future of Work

[40:11–48:03]

• December 2025 marked a tipping point: AI’s coding/agent capabilities jumped from “doing 20% of your tasks to like 80%.”
• OpenAI now sees Codex and agents as tools for everyone, not just coders: “Codex is for everyone.” ([42:36])
• Delegation to agents reframes users as “CEO of a fleet of agents,” but Brockman cautions: “You cannot abdicate responsibility. Accountability stays with the human.” ([45:52])

Model Progression, Tool Use, and Creativity

[48:03–51:38]

• Next steps include deepening tool use (AI using desktops, running businesses), seamless natural-language interaction, and a push toward AI-enabled creativity (science, arts, etc.).
• True creative breakthroughs (like AlphaGo’s “move 37”) may require new forms of training and evaluation, but Brockman is optimistic it’s coming across all fields.

Scaling Compute & Infrastructure Bets

[52:44–62:03]

• Investment in compute and data centers is not a “YOLO” bet but a necessity for continued progress and meeting exploding demand.
• Compute is viewed as a revenue center: “No matter how much we try to build, I know we're not going to be able to keep up with the demand.” ([00:00], [58:08])
• The shift to knowledge work (enterprise and individual productivity) will drive the next wave of revenue.

AI’s Societal Impact: Public Perception, Jobs, and Political Stakes

[64:50–71:33]

• Public skepticism about AI and data centers is addressed—Brockman cites health and personal empowerment stories, rebuffs environmental concerns with stats, and calls for more positive storytelling around AI’s real-life benefits.
• He acknowledges giving political donations to both Republican and bipartisan groups, describing himself as a one-issue donor focused on AI’s role in US competitiveness.

Preparing for the Future

[73:33–74:23]

• For individuals: Approach AI with curiosity; experiment, iterate, learn to manage and collaborate with agents. AGI will “help humans foster more human connection… and spend more time doing what they want. The question is, well, what do you want?” ([73:33])
• “It is much easier to see what you're going to lose than it is to see what you’re going to gain. But it's worth giving it a fair shot.” ([71:44])

Notable Quotes

• On AGI’s Imminence:
  “I think it's extremely clear that we are going to have AGI within the next couple years… the floor of tasks will just be, for almost any intellectual task… the AI will be able to do that.”
  — Greg Brockman, [00:00], [37:18]

• On Strategic Focus:
  “There's too much opportunity. …Every single different idea, as long as it's kind of mathematically sound, you actually can start getting some pretty good results.”
  — Greg Brockman, [06:40]

• On OpenAI’s Underdog Mentality:
  “We never feel complacent, we always feel like we are the challenger… you can never fixate on your competitors. If you focus on where they are, then you'll be where they are and they'll already have moved.”
  — Greg Brockman, [21:15]

• On Safety and Resilience:
  “We invest a lot in safety, security… We need lots of people to be aware. If this technology is going to come and change everything for everyone, people need to participate in that.”
  — Greg Brockman, [31:39], [35:00]

• On Agents and Human Accountability:
  “If you're trying to build a website and your agent messes it up… it's not really the agent's fault, it's your fault. And so you need to care.”
  — Greg Brockman, [45:52]

• On Compute and Growth:
  “Compute… is not a cost center, but a revenue center. Think of it a little bit like hiring salespeople…”
  — Greg Brockman, [56:57]

• On Embracing AI:
  “Try the tools. Because to really understand what it can do for you… only by experiencing the AIs that exist now will that really hit home.”
  — Greg Brockman, [71:44]

Essential Timestamps

• Strategic shift and super app vision: [01:02]–[16:52]
• Tech bets: GPT vs. world/model Sora: [04:44]–[11:19]
• Competition, company culture: [18:35]–[22:59]
• SPUD model and model advances: [22:59]–[28:14]
• Automated researcher/Takeoff: [28:14]–[31:39]
• AGI definitions and timeline: [36:00]–[38:00]
• Coding agents for everyone: [40:11]–[48:03]
• Scaling compute and infrastructure bets: [52:44]–[62:03]
• Public perception, AI’s brand, and politics: [64:50]–[71:33]
• Personal advice for the future: [73:33]–[74:23]

Memorable Moments

• Holiday Party Reflection:
  After ChatGPT’s launch, “the scariest moment at OpenAI was actually after we launched ChatGPT...just feeling this vibe of, we won… I have never felt that. I was like, no, we are the underdog and we always have been.” ([21:15])

• Super App as Iceberg:
  “But what to me is actually much more important is the technological unification. The thing that's really changed… is that it's no longer just about the model, it's about the harness...” ([13:58])

• Unlocking Human Potential:
  “AI is going to free up so much time to increase human connection, to build more bonds… that's not really what being a human’s about. Being a human is about being here, being present, connecting with other humans.” ([51:55])

• Emotional Payoff Examples:
  Brockman shares stories of ChatGPT being used to obtain life-saving medical interventions, highlighting the everyday, personal value AI already provides. ([65:11])

Takeaways for Listeners

• OpenAI is betting on a unified, agentic, reasoning-driven form of AI, pivoting away from running too many parallel, compute-hungry projects.
• The “super app” is designed to democratize both productivity and creativity—serving as your assistant, developer, researcher, and more.
• AGI is viewed as a near-term inevitability; OpenAI is urgently focused on making that transition safe, generative, and beneficial for users and society.
• Success with AI—in both adoption and public perception—requires tangible, life-improving results and clear communication of those benefits.
• For individuals, engaging directly with AI tools is the surest path to understanding—and shaping—what comes next.