May 10, 2024 · 1h 43m · allin
Sam Altman: Getting Fired (and Re-Hired) by OpenAI, Agents, AI Copyright issues
Sam Altman · 34m spoken Jason Calacanis · 19m spoken David Friedberg · 14m spoken Chamath Palihapitiya · 12m spoken David Sacks · 8m spoken
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In Episode 178 of the All-In Podcast, hosts Jason Calacanis, Chamath Palihapitiya, David Sacks, and David Friedberg interview OpenAI CEO Sam Altman regarding OpenAI's product strategy, open-source competition, AI regulation, and his dramatic 2023 board firing. Following the interview, the panel debriefs on AI market dynamics, debates Apple's recent strategic moves, and presents a deep dive into Google DeepMind's AlphaFold 3 breakthrough in computational biology.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The hosts hold 60.4% of the talking time here. How this is scored →
The hosts as informed peer 3.3 Guest teaching 1.9 Guest disagreement 1.2 The hosts pushing back 2.4
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 10:11 Defending closed-source mission reframe
Sam rejects Jason's cynical capitalistic framing of OpenAI's transition from open to closed source, arguing that putting free ChatGPT tools into millions of hands teaches people how to fish and provides massive societal value.
Hardest push from the hosts ▶ 36:56 Challenging AI copyright framingFriedberg directly refuses Sam's assertion that style-based AI generation requires new artist compensation frameworks, arguing that human artists similarly learn from listening to song structures and melodies without paying royalties.
Biggest teaching moment ▶ 12:32 Reframing data dependencies in AI modelsSam corrects Chamath's premise that AI development will degenerate into a proprietary data arms race, explaining that as models become sufficiently intelligent, raw data volume becomes unnecessary for core model training.
The host holds their own ▶ 45:08 Demonstrating expertise on regulatory code auditsFriedberg demonstrates detailed technical and policy expertise by critiquing proposed AI legislation in California, explaining why government source code and weight auditing is fundamentally impractical for dynamic AI models.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| Sam Altman's Career Milestones and OpenAI's Growth | 4 | 2 | 1 | 2 | Sacks and Friedberg probe GPT-5's release timeline and continuous retraining architecture. Sam maintains a measured tone while explaining OpenAI's deliberate release strategy and commitment to making advanced models accessible to free users. | |
| Open Source vs Closed Source Dynamics and OpenAI's Philosophy | 5 | 3 | 2 | 4 | Chamath and Sacks cite specific open-source benchmarks like OpenDevin and Llama 3 to question OpenAI's closed-source approach. Jason presses Sam on whether shifting from open to closed source was driven by commercial motives. | |
| Compute Bottlenecks, Data Arms Race, and AI Infrastructure | 6 | 3 | 2 | 3 | Chamath articulates compute supply chain realities, energy bottlenecks, and data limits. Sam reframes the issue by explaining that sufficiently intelligent models will rely more on algorithmic efficiency than endless training data. | |
| Next-Generation AI Hardware Form Factors and Voice Interfaces | 4 | 2 | 1 | 2 | The hosts inquire about new AI hardware form factors and voice interface latency. Sam notes that the high utility of smartphones sets an extremely high bar for any new consumer device category. | |
| Multimodal Vision, Assistant Philosophy, and Agentic App Interfaces | 5 | 4 | 2 | 3 | Sacks and Chamath explore how AI assistants might replace traditional app user interfaces. Sam educates the hosts on why he prefers an independent senior employee paradigm that pushes back over an automated sycophant. | |
| Transformative AI Applications: Education, Healthcare, and Scientific Discovery | 6 | 4 | 1 | 3 | Friedberg details protein modeling and complex biological networks. Sam clarifies that developing generalized reasoning capabilities will unlock scientific domains faster than building narrow, specialized models from scratch. | |
| AI Copyright, Fair Use, Artist Compensation, and Model Spec | 7 | 3 | 3 | 6 | Friedberg directly challenges Sam's distinction between AI style replication and human musicians learning musical structures. Jason pushes on fair use standards and intellectual property rights for content creators. | |
| AI Regulation, Government Audits, and Regulatory Capture | 7 | 3 | 3 | 6 | Friedberg and Jason push back strongly against state-level regulations and government code auditing proposals. Sam clarifies his focus on international safety testing for frontier models capable of severe global harm. | |
| Universal Basic Income (UBI) Study and Universal Basic Compute | 4 | 3 | 1 | 2 | The hosts ask about Sam's five-year Y Combinator UBI study. Sam shares insights on direct cash transfers and introduces the concept of Universal Basic Compute as an alternative resource allocation model. | |
| Inside the OpenAI Board Firing and Re-Hiring Drama | 4 | 2 | 2 | 5 | Chamath and Jason press Sam for details surrounding his sudden firing from OpenAI while in Las Vegas. Sam recounts the chaotic sequence of board communications and rapid leadership changes. | |
| Non-Profit Board Dynamics and Culture Clash | 5 | 3 | 2 | 4 | Sacks asks whether a culture clash between non-profit directors and commercial startup leaders caused the board rupture. Sam defends the board members' safety intentions while acknowledging disagreement with their actions. | |
| Sam Altman on OpenAI Equity, Motivation, and Structural Rumors | 5 | 3 | 3 | 5 | Chamath and Jason question why Sam holds no equity in OpenAI and address rumors about side projects. Sam candidly states he regrets not taking equity solely because the decision fueled persistent conspiracy theories. | |
| Clarifying the $7 Trillion AI Infrastructure Rumor and OpenAI's Organizational Model | 4 | 2 | 1 | 3 | Jason asks about reports that Sam was raising $7 trillion for AI hardware infrastructure. Sam dispels the figure as inaccurate while emphasizing the real need for massive global compute investment before departing. | |
| Host Debrief: AI Economics, Board Governance, and Executive AI Agents | 0 | 0 | 0 | 0 | The hosts conduct a post-interview debrief on AI infrastructure economics, model scaffolding, and corporate governance without the guest present. | |
| All-In Summit Announcements, Tequila Launch, and Campus Protest Debates | 0 | 0 | 0 | 0 | The hosts discuss event planning, brand launches, and legal aspects of First Amendment speech during campus protests in a host-only segment. | |
| Critiquing Apple's Controversial 'Crush!' iPad Pro Advertisement | 0 | 0 | 0 | 0 | The hosts analyze Apple's controversial iPad advertisement, M&A conservatism, and capital allocation in a host-only segment. | |
| David Friedberg Explains Google AlphaFold 3 Breakthrough | 0 | 0 | 0 | 0 | Friedberg delivers a monologue explaining Google AlphaFold 3's capabilities in modeling molecular interactions for biological discovery. | |
| Simulating Biological Aging and Disease Treatments in Silico | 0 | 0 | 0 | 0 | Friedberg explains in silico simulation of cellular de-aging and Yamanaka factors in a host-only monologue. | |
| Banter on Google's AI Business Strategy and Closed-Source Models | 0 | 0 | 0 | 0 | The hosts discuss Google's commercial monetization strategy for proprietary biological AI models in a host-only banter segment. | |
| Episode Outro Song and Audio Visualizer | 0 | 0 | 0 | 0 | This segment consists of the episode outro, closing banter, house credits, and social media plugs. |
Statements from this episode (44)
Calacanis: Sequoia's first Scout fund is its highest multiple returning fund
“That tiny experimental fund that Sam and I were part of a scouts is Sequoia's highest multiple returning fund.”
Altman: AI development will shift toward continuous updates over major discrete releases
“I think I think that's like a better hint of what the world looks like, where it's not the like one, two, three, four, five, six, seven, but you just, you use an AI system, and the whole system just gets better and better fairly continuously. I think that's li…”
Chamath: Open-source OpenDevin is almost as good as Cognition's Devin
“And then some kid just published it under an open MIT license, like open Devon. And it's incredibly good and almost as good as that other thing that was closed source.”
Altman: Advanced AI will eventually become instantaneous and too cheap to meter
“I don't know, like, when we get to intelligence too cheap to meter and so fast that it feels instantaneous to us and everything else, but I do believe we can get there for, you know, a pretty high level of intelligence, and I, it's important to us, it's clearl…”
Altman: OpenAI aims to build a general intelligence layer, not just models
“What we're trying to do is not make the sort of smartest Set of weights that we can, but we're trying to make is like this useful intelligence layer for people to use. And a model is part of that.”
Altman: AI will not become an arms race for training data
“I definitely don't think it'll be an arms race for data because when the models get smart enough at some point, it shouldn't be about more data, at least not for training.”
Altman: OpenAI will achieve major algorithmic efficiency gains
“Well, we'll, we'll make huge algorithmic gains for sure, and I don't want to discount that.”
Altman: Energy supply is a major bottleneck for AI scaling
“There's finding the energy, which is a huge bottleneck.”
Altman: The iPhone is humanity's greatest piece of technology
“I personally think an iPhone is, like, the greatest Piece of technology humanity has ever made.”
Altman doubts standalone AI hardware devices can succeed against smartphones
“So if you could like make a way cheaper device, I think the barrier to carry a second thing or use a second thing is pretty high. So I don't think, given that we're all willing to pay for phones, or most of us are, I don't think cheaper is the answer.”
Sam Altman predicts smartphone manufacturers will add specialized AI chips
“Probably, but the phone manufacturers are going to do that for sure.”
Altman: AI assistants will evolve as separate entities, not personal alter egos
“I personally like the separate entity approach better and think that's where we're going to head. And so in that sense, the thing is not you, but it's like a always available, always great, super capable assistant executive.”
Altman: Humanoid robots are superior because the physical world is designed for humans
“Same reason I'm like more excited about humanoid robots than sort of robots of like very other shapes. The world is very much designed for humans, and I think we should absolutely keep it that way.”
Altman: Voice AI will not completely eliminate visual screen interfaces
“I think visual user interfaces are super good for a lot of things. And it's hard for me to imagine like a world where you never look at a screen and just use voice mode only, but I can imagine that for a lot of things.”
Altman: GPT-4 cannot yet drive major scientific discoveries independently
“GPT-IV clearly not there in a big way, although maybe it accelerates things a little bit by making scientists more productive.”
Altman: Solving general AI reasoning will allow easy adaptation to specialized domains
“You clearly will need specialized simulators, connectors, pieces of data, whatever, but my intuition, and again, I don't have this like backed up with science, my intuition would be if we can figure out the core of generalized reasoning, connecting that to new…”
Altman: OpenAI's Sora video model was built from scratch without a language model base
“But yeah, Sora, like, does not start with a language model. It's, that, that's a model that is, like, customized to do video and so, like, we're clearly not at that world yet.”
Altman: Best text models are autoregressive while best video models use diffusion
“As far as I know, all the best text models in the world are still autoregressive models and the best image and video models are diffusion models.”
Altman: AI copyright debates will shift from training data to inference time
“I think the conversation has been historically very caught up on training data, but it will increasingly become more about what happens at inference time as training data becomes less valuable”
Altman: OpenAI decided against entering music AI due to copyright concerns
“We have currently made the decision not to do music, and partly because exactly these questions of where you draw the lines”
Sam Altman: Frontier AI will eventually pose significant global harm risks
“I think there will come a time in the not super distant future, like, you know, we're not talking like decades and decades from now, where AI says the frontier AI systems are capable of causing significant Global harm.”
David Friedberg: Proposed AI bills require pre-deployment government audits
“Basically, the California legislation that's proposed and some of the federal legislation that's been proposed Basically requires the federal, the government to audit a model, to audit software, to audit and review the parameters and the weightings of the mode…”
Altman: Statutory AI laws will be obsolete within 12 to 24 months
“In 12 months, it will all be written wrong, and I don't think, even if these people were, like, true world experts, I don't think they could get it right, looking at it in 12 or 24 months.”
Altman: Results from five-year YC UBI study will be published soon
“Our results in that come out very soon. I just, it was a five-year study that wrapped up or started five years ago.”
Altman: Future social safety nets may provide universal basic compute over income
“I wonder if the future looks something like more like universal basic compute than universal basic income. And everybody gets like a slice of GPT sevens compute and they can use it. They can resell it. They can donate it to somebody to use for cancer research.…”
Altman: Emmett Shear was great during OpenAI's interim CEO transition
“Emmett's great. And I have nothing but good things to say about Emmett in that whole process.”
Altman: OpenAI's board called him just before tweeting news of his firing
“They did call me first before tweeting, which was nice of them.”
Altman: I never doubted OpenAI board's integrity despite my firing
“Not all of those board members are my favorite people in the world, but I have serious respect for the gravity with which they treat AGI and the importance of getting AI safety right, and even if I stringently disagree with their decision making and actions, w…”
Altman: AGI is unavoidable and will be tremendously beneficial
“I think it is unavoidable that this is going to happen. I also think it's going to be tremendously beneficial, but we do have to navigate how to get there in a reasonable way.”
Altman: Non-profit board rules led to initial decision against OpenAI equity
“The decision back then, the re, the original reason was just like the structure of our nonprofit. It was like there was something about Yeah, ok, this is like, nice from a motivations perspective, but mostly it was that our board needed to be a majority of dis…”
Altman: Device and chip ventures belong to OpenAI, not him personally
“So the things like, you know, device companies, or if we were doing some chip fab company, it's like, those are not Sam Project. Those would be like opening. I would get that equity.”
Altman denies knowing origin of reported $7 trillion AI chip figure
“I don't know where that came from, actually. I genuinely don't.”
Altman: World needs far more AI infrastructure than currently planned
“I think the world needs a lot more AI infrastructure, a lot more than it's currently planning to build, and with a different cost structure.”
Altman: Concentrating entire company on single AI models works best
“These systems are so complicated and concentrating bets are so important. Like one You know, at the time, before it was, like, obvious to do this, you have, like, DeepMind or whatever has all these different teams doing all these different things, and they're …”
Palihapitiya: OpenAI will be one of four dominant AI companies
“I think that these guys are going to be one of the four major companies that matter in this whole space.”
Palihapitiya: AI models will commoditize while scaffolding and interfaces capture value
“So I think the right way to think about this now is the models will basically be all really good. And then it's all this other stuff that you'll have to pay for interface. Whoever builds all this other stuff. Is going to be in a position to build a really good…”
Sacks: There is no innovation left in hardware devices like iPads
“I mean, there's no innovation left, so they're just making the devices, you know, thinner.”
Sacks: Apple's iPad 'Crush!' ad was off-brand for destroying creative tools
“I think that by destroying all those creative tools that Apple is supposed to represent, it definitely seemed very off brand for them.”
Chamath predicts Apple will use excess capital primarily for stock buybacks
“They're probably just going to buy back the stock.”
Friedberg: Apple creates value by defining entirely new product categories
“Where they've really created value is on how unique the products are. They almost create new categories.”
Friedberg: Google's AlphaFold 3 is a breathtaking moment for biology and medicine
“Alpha fold three that was just announced today and demonstrated by Google is a real I would say breathtaking moment for biology, for bioengineering, for human health, for medicine.”
Friedberg: Google kept AlphaFold 3 proprietary instead of open-sourcing the code
“They've basically kept all the IP for AlphaFold III in Isomorphic. So Google is going to monetize the heck out of this capability. And what they made available was not open source code, but a web-based viewer that scientists for, quote, non-commercial purposes…”
Friedberg: AlphaFold 3 predicts molecular binding to specific DNA sequences
“One of the things that this alpha fold three can do is predict what molecules will bind and promote certain sequences of DNA, which is exactly what we try and do with the Yamanaka factor based expression systems and find ones that won't trigger off target expr…”