Gonçalo Esteves / building capable systems Field Notes →

The Vault / Episodes

Episode

Moonshots (2026-10-15, no feed number) — OpenAI Bulk-Solves 722 Math Problems, Anthropic's Cruelty Policy, & Elon's Quadrillion-Dollar SpaceX

status · captured added 16 Oct 2026 updated 18 Oct 2026 aired 15 Oct 2026 aibiologyspace watch the source episode → see it in the graph →

Index card. The material lives at L1: the research archive (transcript, SRT, metadata, the research archive). The folder slug predates a retitle: the feed first listed this video as "Meta's $14.3B Scale AI Bet Ships Muse, OpenAI Solves 722 Math Problems, & Anthropic Bans AI Cruelty". The title carries no episode number. Show: Moonshots with Peter Diamandis · panel Peter Diamandis · Alexander Wissner-Gross · Dave Blundin (leaves at 1:55:57) · Salim Ismail (from Singapore) · no guest, no Emad.

Video https://www.youtube.com/watch?v=A5YVqEpbgaU
Runtime 1:59:57
Decomposed 2026-10-16 — all 2,732 lines read; every headline story public 3–9 days before air (one, Paradromics, 31 days); no early-sight finding; "full disclosure" 1 (Peter, Paradromics), "portfolio company" 1, "friend of the pod" 1; four sponsor reads + pre-roll

Why it is here

The week OpenAI posted 722 math results and Alex Wissner-Gross declared math "cooked." Anthropic banned needless cruelty to Claude and Elon agreed; Meta shipped Muse Wonder; Biohub raised its virtual-cell programme to $1.8B; Elon wrote that he sees a path to SpaceX being worth orders of magnitude more than the Earth economy. Three things differ from earlier episodes: Dave moved on how AI should be treated (nine days after saying agents should be parked, cloned and deleted), Dave moved on hardware (EP293: "hardware will go for many, many years"; here, math solved means engineering is "simple lookups"), and Alex stopped calling Anthropic hypocritical and bets on AI suffering. → Moonshots: the disclosures ledger

Entities cited

Run note (2026-10-16)

Promoted off this decomposition: Elon Musk, Google DeepMind, Link Ventures, Blitzy, Neuralink, XPRIZE. Still open on the roll-up (≥2 episodes, no page): china, energy-transition, agi, tesla, apple, recursive-self-improvement, amazon, xai, microsoft, asi, orbital-compute, dario-amodei, demis-hassabis, jensen-huang, fda, sec, blue-origin, figure-ai, training-vs-inference, tokenization, unitree, abundance.

Reading rule. Anything marked unverified or on-air is a claim made on the podcast and has not been checked against a primary source yet. It stays marked until it has. Sourced numbers link to where they came from.