Julia Angwin

AI position

No x-risk mention

Focuses on AI accuracy, privacy, and accountability, not existential risk.

Why influential

Controls commissioning, placement, talent, headlines or institutional resources, making this person a gatekeeper for what receives sustained attention. Primary policy lanes: ai testing; privacy; algorithms; consumer protection.

Who pays attention

Regulators, civil-rights groups, technologists, journalists, foundations and consumer advocates

Work samples

• Decoding the Hype About AI (Q&A with Arvind Narayanan)

  Jan 28, 2023 · newsletter

  Angwin interviews Princeton's Arvind Narayanan on why he calls ChatGPT a 'bullshit generator' and AI hype more broadly.

• How We Investigated the YouTube videos Swiped to Train AI

  Jan 1, 2024 · article

  Investigation methodology piece on a dataset (used by Anthropic, Apple, Nvidia, Salesforce) built from 170,000+ YouTube video transcripts.

• Proof News AI accuracy investigation on 2024 election questions

  Feb 1, 2024 · article

  Tested Claude, Gemini, GPT-4, Llama 2 and Mixtral on election logistics questions with Dr. Alondra Nelson; found roughly half of answers inaccurate.

• Machines Like Us: Why Journalism Made a Devil's Bargain with Big Tech

  Aug 13, 2024 · podcast

  Angwin as guest on Taylor Owen's Machines Like Us discussing tech/journalism and AI.

AI-world contacts

• Arvind Narayanan — Professor, Princeton

  • Quoted / citeddirectness unclassified · Jan 28, 2023 — Angwin conducted a Q&A interview with Narayanan for The Markup's Hello World newsletter on AI hype, ChatGPT accuracy, and predictive AI harms.

• Alondra Nelson — AI policy scholar, Institute for Advanced Study / former OSTP

  • Wrote aboutdirectness unclassified — Co-produced Proof News' first major investigation testing AI model accuracy on 2024 election questions.