postgrok

ai-doom-paper-trail

Properties1
tagsai, sources, citation, ai-harms, effective-altruism

9 min read

The AI doom debate’s paper trail

Definition: the primary sources behind the name-drops in 2026-10-04-youtube-newport-ai-harm-taxonomy. Cal Newport’s video “An Insider’s Guide to AI Doom” cites roughly a dozen works, people, and claims casually, mostly without titles or links. A listener has to use context clues to work out which source Newport means in each case. This page does that digging: for every name-drop it names the primary source, gives the exact record, and grades how well Newport’s telling matches it. A web-search pass on 2026-10-04, eight searches, produced every entry. Each entry carries its own provenance line so the next reader can re-check it.

Doomers

The TESCREAL acronym

Newport credits the term to “Timnit Gebru and collaborators” (21:30). The paper: Timnit Gebru and Émile P. Torres, “The TESCREAL bundle: Eugenics and the promise of utopia through artificial general intelligence”, First Monday 29(4), published April 14, 2024, doi:10.5210/fm.v29i4.13636, https://firstmonday.org/ojs/index.php/fm/article/view/13636. The caption garble “Emad Mostaque” at 21:50 resolves to Torres; the vault corrected this on ingest day. See tescreal. Grade: exact match, and the acronym’s letters as Newport reads them match the paper’s list.

The Yudkowsky lineage

Newport traces Eliezer Yudkowsky from the Extropian mailing lists, through Rationalism, to “his own techno-utopian institute” and the conviction that superintelligence is inevitable and could just as easily kill everyone (3:48 to 4:32). The records, beat by beat:

  • The Extropians mailing list ran 1996 to 2003, operated by the Extropy Institute. Yudkowsky is a documented participant. Searchable archive: https://extropians.boydkane.com, raw data at macterra/extropians on GitHub.
  • LessWrong was founded February 2009 from Yudkowsky’s posts on Overcoming Bias (2006 to 2009, principal contributors Yudkowsky and Robin Hanson). The site’s own history page records that early users were “recruited via Eliezer’s transhumanist SL4 mailing list”: https://www.lesswrong.com/w/history-of-less-wrong.
  • The institute is MIRI, the Machine Intelligence Research Institute, originally the Singularity Institute.
  • The conviction is the book: Yudkowsky and Nate Soares, If Anyone Builds It, Everyone Dies, Little, Brown, September 16, 2025.

Grade: the lineage holds. Each of Newport’s three beats maps onto a documented artifact.

The doomer civil war

Newport’s chapter “Anthropic, OpenAI, and the doomer civil war” (9:12) says external rationalists attack the labs for moving too fast. The primary record is Yudkowsky’s own output: the 2025 book above argues any lab building the technology dooms everyone, and his post “Only Law Can Prevent Extinction” (LessWrong, April 2026) calls for government shutdown of the effort. Grade: real and on the record, hostile to the labs in their founders’ own tradition.

The Amodei group house

Newport, unnamed and second-hand (10:52 to 11:07): Dario Amodei “used to live in a group house with other really major effective altruists. His sister, who helped start Anthropic, was married one of them.” The referents are named in the Wall Street Journal’s Amodei reporting (August 2026): “Amodei, his friend Holden Karnofsky and his sister, Daniela—who would later marry Karnofsky—lived with other key figures from the AI safety and Effective Altruism” world. https://www.wsj.com/tech/ai/claude-dario-amodei-wife-anthropic-e1eeda7d. Karnofsky co-founded GiveWell and leads Open Philanthropy (general reference; the WSJ piece supplies the house and the marriage). Grade: confirmed, including the marriage. The unnamed spouse is Holden Karnofsky.

The Bankman-Fried money

Newport mentions Sam Bankman-Fried as Doomer funding at 11:10. Primary: New York Times, “How the Collapse of Sam Bankman-Fried’s Crypto Empire Has Disrupted A.I.”, December 1, 2022: a round led by Bankman-Fried’s FTX had fed “at least 255.8 million of FTX-era giving to organizations fighting potential risks from advanced AI. The stake’s size is unresolved across outlets: Benzinga (September 2025) says 2.5 billion valuation, an 8% stake; Business Insider (August 31, 2026) reads court records for 13.56% of the 2022 Series B. Grade: the money is confirmed, the percentage is not.

The EA bookshelf

Newport names MacAskill, Ord, and Bostrom without works. The standard primary texts: Bostrom, Superintelligence (Oxford University Press, 2014); Ord, The Precipice (2020); MacAskill, What We Owe the Future (2022). This pass verified MacAskill’s movement role, not the three books themselves: the NYT piece above calls him a founder of the Center for Effective Altruism and a lead of the FTX Future Fund, and Fortune (November 15, 2022) reports that MacAskill pitched Bankman-Fried in 2012 on making money specifically to donate it. The three titles are filed here as general reference, marked so on purpose.

Accelerationists

The Andreessen text

Newport’s VCs-then-White-House composition (14:03 chapter) has a founding document. Primary: Marc Andreessen, “The Techno-Optimist Manifesto”, Andreessen Horowitz, October 16, 2023, https://a16z.com/the-techno-optimist-manifesto/. It states the camp’s creed in its own words: “We believe in accelerationism – the conscious and deliberate propulsion of technological development”, and “any deceleration of AI will cost lives”. Grade: the manifesto is idealist on its face. Newport’s money framing comes from him, not from this trail; his own list at 16:35 (returns, chip sales, export controls, Silicon Valley-White House alliances) is commentary he doesn’t cite either. The text at least confirms the accelerationist creed.

The Jensen Huang pushback

Newport’s paraphrase at 15:45 (“what’s with all this sci-fi stuff, this is crazy, just don’t build unsafe things, this is a fairy tale”) has several recorded originals:

  • TechRadar, June 6, 2026, quoting Huang from January 2026: “I think we’ve done a lot of damage with very well-respected people who have painted a doomer narrative, end of the world narrative, science fiction narrative. And I appreciate that many of us grew up and enjoyed science fiction, but it’s not helpful.”
  • Fortune, May 2, 2026, from the Memos to the President podcast: “Somehow, because they became CEOs, you adopt a God complex”, aimed at Amodei’s warning record.
  • CBS News interview, September 2026: Huang called extinction-by-2030 warnings fiction with zero chance, and said spreading that fear without scientific backing is irresponsible (reporting via Bangkok Post).
  • All-In Summit, Los Angeles, September 14, 2026 (Business Insider): on doomer talk, “We’re making it up.”

Grade: faithful. The paraphrase compresses at least three separate on-record statements.

The Zuckerberg post

Newport, flagged as his own paraphrase (15:54): Zuckerberg asked “why is Anthropic obsessed with recursive self-improvement” weeks before the video. The post is from mid-September 2026, after Dario Amodei proposed a coordinated slowdown. Actual line, preserved: “Committing the significant majority of compute towards serving people rather than racing towards recursive self-improvement is one of the best ways to ensure we develop this technology safely” (Mercury News, September 16, 2026, https://www.mercurynews.com/2026/09/16/mark-zuckerberg-meta-ai-slowdown). The target: Anthropic’s June 2026 statement defining recursive self-improvement and warning “it could come sooner than most institutions are prepared for” (The Verge, June 4, 2026). Wider corpus, same arc: NYT interview July 28, 2026, and Zuckerberg’s 6,500-word essay of August 10, 2026 (The Atlantic, Bloomberg). Grade: timing matches. Video posted October 1, “a couple weeks ago” fits the September post.

The Hill claim, graded hardest

Newport says congressional Democrats, and some Republicans, are being steered by Doomers, sourced second-hand from Hill staffers, and adds that EA fellowship programs are placing effective altruists onto congressional staff as “a plan” (24:55 to 25:52). Journalism documents the mechanism:

  • Politico, October 13, 2023, “How a billionaire-backed network of AI advisers took over Washington”: Open Philanthropy funded the Horizon Institute for Public Service to place AI fellows on the Hill, where they “are already involved in negotiations that will shape Capitol Hill’s accelerating plans to regulate AI”, drafting bills and educating members. A Google document last updated by Open Phil in August 2022 said the fellowship would consider candidates “who share Open Philanthropy’s interests in global catastrophic risks and technology’s long-term impacts”. Ethics experts (Project On Government Oversight, a former Office of Congressional Ethics director) called it a conflict of interest. Horizon’s director denies the program pursues policy goals. https://www.politico.com/news/2023/10/13/open-philanthropy-funding-ai-policy-00121362
  • Politico, December 3, 2023: a second channel, an AAAS “rapid response cohort” of Hill AI fellows funded by Microsoft, OpenAI, Google, IBM, and Nvidia. https://www.politico.com/news/2023/12/03/congress-ai-fellows-tech-companies-00129701
  • Washington Examiner, November 6, 2025, reports Horizon deployed fellows “at influential think tanks and government offices”, and that some worked on policy affecting EA goals, such as chip export controls.
  • New York Post, March 8, 2024, reports EA-linked placements in the Biden White House and cites $55 million from Open Philanthropy to Georgetown’s CSET center.

Grade: two funding channels are on the record, Horizon’s and the AAAS cohort’s, across four outlets. Newport’s telling exceeds the record in three ways. The “this is a plan” framing goes beyond any report, and Horizon denies intent. The strongest reporting dates to 2023, three years older than his present-tense “I’m hearing this directly”. And his staffer-conversation sourcing can’t be checked by anyone who isn’t on the Hill. See horizon-institute and open-philanthropy.

The Centrists’ own shelf

NTCS, “powerful but normal”

Newport names Arvind Narayanan (22:26) inside the “normal technology computer scientists” subgroup. The subgroup’s book: Arvind Narayanan and Sayash Kapoor, AI Snake Oil: What Artificial Intelligence Can Do, What It Can’t, and How to Tell the Difference, Princeton University Press, September 24, 2024. Publisher description, preserved: the book explains “why AI isn’t an existential risk, and why we should be far more worried about what people will do with AI than about anything AI will do on its own” (https://press.princeton.edu/books/hardcover/9780691249131/ai-snake-oil). Narayanan’s Princeton homepage carries the framing in three words: “AI as normal technology”. Grade: this is the citation Newport gestures at without naming. See arvind-narayanan.

The AI ethics shelf

Pre-LLM touchstone for the first Centrist subgroup: Emily Bender, Timnit Gebru, Angelina McMillan-Major, and Margaret Mitchell, “On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?”, FAccT 2021, doi:10.1145/3442188.3445922, https://dl.acm.org/doi/10.1145/3442188.3445922. Gebru’s firing from Google over it: New York Times, December 3, 2020.

Sources

  • The name-drops graded here: 2026-10-04-youtube-newport-ai-harm-taxonomy, timestamps 3:48 to 26:39, from auto-generated captions.
  • Every URL and record above: web-search deep-dive of 2026-10-04, eight searches, agent-run.
  • Grades are the agent’s reading of the sources. Newport doesn’t grade anything on air; he marks his own paraphrases and leaves it there.

Graph View

Backlinks