The four besties — Jason Calacanis, Chamath Palihapitiya, David Sacks, and David Friedberg — take up reporting that Anthropic has been asking whether Claude might be conscious, OpenAI’s latest math release, unrest and yields in France, and a close on agents. Consciousness claims stay unproven. Companies and markets named here were mentioned on the show. Nothing here is investment advice.
Chapters
- 0:00 The original quartet is back on the desk.
- 2:53 Anthropic, Claude, and a consciousness question.
- 33:38 Yudkowsky, LessWrong, and Roko’s basilisk.
- 38:28 OpenAI’s math release, and a backlash.
- 1:02:03 Riots in France, a socialism metric, and the bond market.
- 1:21:41 Agents, a headless Grok Bot, and whether IP still matters.
Show more
Want this section to go deeper? Share this page with the creators and have them sign up so we can get permission to dig deeper into this episode.
Key topics
- 0:00 The original quartet is back on the desk — Jason Calacanis opens with all four besties on the desk: Chamath Palihapitiya (he is in Tokyo, in his telling, and ribs Chamath about having to book a suite), David Sacks, and David Friedberg. Companies and markets named later were mentioned on the show. Nothing here is investment advice.
- 2:53 Anthropic, Claude, and a consciousness question — Jason tees up New York Times reporting that Anthropic has been hosting religious leaders and philosophers — including outreach described as lobbying the Pope — on whether Claude might be conscious, plus a ‘model welfare’ conversation and alignment risk. The desk argues over whether this is a serious safety question, a new kind of religion, or a dangerous idea (Sacks compares it with Asimov’s Three Laws; Friedberg and Chamath push on what ‘defiant’ models would mean; Dario Amodei’s writing on superintelligence comes up). Consciousness claims here are the reporting and the besties’ debate, unproven. Nothing here is investment advice.
- 33:38 Yudkowsky, LessWrong, and Roko’s basilisk — As a possible explanation for the vibe, one of the hosts walks through Eliezer Yudkowsky, the LessWrong board, Roko’s basilisk, and the doom-preaching tradition around superintelligence, including the book If Anyone Builds It, Everyone Dies. That history is their recap, not an endorsement of the argument. Mentioned on the show.
- 38:28 OpenAI’s math release, and a backlash — Jason turns to OpenAI putting out a large set of math papers and results (hundreds of documents, in the numbers they use) from an unreleased model, after an earlier Millennium Prize claim the desk had discussed. Friedberg walks through why some mathematicians are uneasy about a shift from solving problems to constructing problem sets for AI. Figures and status of the proofs are from the reports they read. Mentioned on the show. Nothing here is investment advice.
- 1:02:03 Riots in France, a socialism metric, and the bond market — The desk asks whether new unrest in France is different from past waves. They cite injuries, a government blame-the-far-left framing, and French 10-year yields they put above 5%, then compare U.S. long-term yields. Friedberg offers a jokey ‘socialism point of guaranteed return’ frame; later talk turns to ‘bond vigilantes’ putting Western governments on notice. Political and market claims are theirs, attributed. Mentioned on the show. Nothing here is investment advice.
- 1:21:41 Agents, a headless Grok Bot, and whether IP still matters — Jason says agents will be the main story. They discuss Elon Musk saying Grok Bot will go ‘headless’ (pick a best end-to-end model, in their telling), practical agent use, an OpenAI product they call Dots, and whether software/IP gets cheaper as tools can rebuild things like a GTA-scale game. Market and product talk is theirs. Mentioned on the show. Nothing here is investment advice.
Show more
Want this section to go deeper? Share this page with the creators and have them sign up so we can get permission to dig deeper into this episode.
People
On this episode
Jason Calacanis
Host. Opens the show, tees up the Anthropic and OpenAI blocks, and closes on agents and Grok Bot. 0:00 2:53 1:21:41
Chamath Palihapitiya
Host. On the desk from Tokyo, in his telling. In the consciousness and later market/agent blocks. 0:00 8:09 1:02:03
David Sacks
Host. Pushes on model ‘defiance’ versus Asimov-style rules, and on several later docket items. 0:00 24:14 38:28
David Friedberg
Host. Walks the OpenAI math backlash, then offers his ‘socialism point of guaranteed return’ frame on France and bonds. 0:00 3:01 38:28
People mentioned
Pope Leo XIV
“The Pope” in the New York Times story they open on: Anthropic’s outreach to religious leaders, and the Pope’s first encyclical, on AI. 3:18 4:17
Isaac Asimov
His Three Laws of Robotics come up as a contrast to Anthropic’s “defiant” model idea. 24:26
Dario Amodei
Cited in the consciousness/alignment block for writing about superintelligence that should be ‘loving’ and protective of humanity, in their telling. His writing, their commentary. 27:01
Eliezer Yudkowsky
Described as a central AI-doom voice around LessWrong, Roko’s basilisk, and the book If Anyone Builds It, Everyone Dies. Their recap, not an endorsement. 33:38
Elon Musk
Comes up on housing talk and, later, a tweet they recap about making Grok Bot headless. Their recap. Nothing here is investment advice. 1:11:33 1:22:07
Show more
Want this section to go deeper? Share this page with the creators and have them sign up so we can get permission to dig deeper into this episode.
Watch & read
Watch on YouTube — Full episode on the All-In Podcast YouTube channel. 0:00 · YouTube 0:00
All-In official site — The show’s official site. 0:00 · allin.com 0:00
Show more
Want this section to go deeper? Share this page with the creators and have them sign up so we can get permission to dig deeper into this episode.
Products
Anthropic / Claude — The consciousness, model-welfare, and lobbying story, as they recap the Times reporting. Unproven as a claim about sentience. Mentioned on the show. Nothing here is investment advice. · Wikipedia · Wikipedia 2:53
OpenAI — The math-paper release and, later, a product they call Dots, plus agent talk. Figures are from reports they read, not an audit. Mentioned on the show. Nothing here is investment advice. · Wikipedia 38:28
Show more
Want this section to go deeper? Share this page with the creators and have them sign up so we can get permission to dig deeper into this episode.
Go Deeper: rabbit holes
Rabbit holes from this episode — people, cases, companies, and works they actually discussed. Companies and markets named on the show were mentioned on the show. Nothing here is investment advice.
- 3:01 The New York Times reporting they start from — Jason’s tee-up is Times reporting on Anthropic sessions with religious leaders and philosophers under NDAs, including outreach described as involving the Pope. The article is their starting point.
- 24:14 Asimov’s Three Laws, as a contrast — Sacks and the desk compare Anthropic’s ‘defiant’ model idea with Isaac Asimov’s Three Laws of Robotics — in their telling, almost the opposite instinct. Wikipedia · Wikipedia
- 33:38 Roko’s basilisk, and If Anyone Builds It, Everyone Dies — The thought experiment and Yudkowsky’s book, as they recap that history. Not an endorsement. Wikipedia · Wikipedia · Wikipedia
- 1:02:03 French yields, and the ‘bond vigilantes’ frame — They put France’s 10-year above 5% and talk about Western debt coming under pressure. Their market talk, as of this taping. Nothing here is investment advice. Wikipedia
Show more
Want this section to go deeper? Share this page with the creators and have them sign up so we can get permission to dig deeper into this episode.
Show plugs
More from this show — allin.com · All-In on YouTube
Show notes are a summary, not a full transcript. As an Amazon Associate we earn from qualifying purchases. Links on this page may be affiliate links through Amazon and other partner programs we participate in now or in the future. If you buy through those links, we may earn a commission at no extra cost to you.