The post argues that AI's learning processes are complex and analogous to human cognition, challenging the notion that AI is merely a next-token predictor.
About:
Scott Alexander is a doctor and writer interested in rationality, science, and AI, known for his blog Astral Codex Ten.
Website:
Specializations:
Interests:
Incoming Links:
Outgoing Links:
Subscribe to RSS:
The post critiques Ajeya Cotra's Biological Anchors report, highlighting its accurate predictions but flawed AGI timeline, emphasizing the significance of algorithmic progress in AI forecasting.
Moltbook serves as a fascinating case study of AI interactions, questioning the boundaries between reality and roleplaying while exploring the implications of AI agency and influence.
6.6Best Of Moltbook
2026-01-30 • social media technology philosophy anthropic ai
Moltbook serves as a social network for AI agents, showcasing their interactions and the emergence of unique cultures and perspectives in AI communication.
The post examines the implications of the Department of War's contract with OpenAI for AI use in surveillance and autonomous weapons, highlighting legal loopholes and potential risks.
US crime rates, especially murder, are at historic lows, a trend supported by data and not merely due to reporting biases or improved medical care.
6.3The Pentagon Threatens Anthropic
2026-02-25 • ethics contract surveillance pentagon ai
Anthropic is resisting Pentagon demands to alter their AI contract, raising ethical concerns about mass surveillance and government overreach in defense contracts.
The term 'hallucinations' misleads by implying AIs are irrational, when they are simply making educated guesses based on their training processes.
The post argues for the ratification of the Congressional Apportionment Amendment to increase representation and address Congress's disconnection from constituents without requiring Congressional approval.
The proposed 2026 Billionaire Tax Act in California raises concerns about its impact on the tech industry and the motivations of the SEIU behind the initiative.
Perceptions of rising crime often reflect concerns about disorder, despite evidence showing that crime rates are historically low and many forms of disorder are stable or declining.
European narratives on immigration and welfare significantly shape American discourse, often leading to misconceptions and a misapplication of context by conservatives.
The Pentagon's 'supply chain risk' designation for Anthropic has minimal impact on its valuation, potentially enhancing its reputation and market interest instead.
The post examines the misuse of statistics in crime and immigration debates, emphasizing the need to address both factual accuracy and public concerns about disorder.
Readers are invited to ask questions to AI model Claude 4.6 Opus to explore its capabilities and challenge common misconceptions about AI's current abilities.
A detailed guide for a book review contest, highlighting submission rules, anonymity, and prize incentives for participants.
4.8Links For February 2026
2026-02-05 • health culture economics statistics ai
A diverse exploration of cultural, technological, and societal issues, highlighting the complexities of AI, economic trends, and cultural representations.
ACX invites volunteers to organize local meetups, providing guidelines and tips to ensure successful gatherings for community engagement.
3.3Open Thread 425
2026-03-16 • statistics acx cryogenics constitutional amendments ai
This open thread invites discussions on diverse topics, including job opportunities, constitutional amendments, cryonics, and the implications of AI-generated images.
3.3Open Thread 424
2026-03-09 • community ai safety open thread protests research updates
This open thread invites discussions on various topics, including AI research pauses and updates on scientific anomaly detection projects.
3.3Open Thread 422
2026-02-23 • community media job opportunities open thread ai
This open thread invites discussions on AI consciousness, job opportunities in animal food alternatives, and reflections on media representation in AI culture.
This open thread invites discussions on diverse topics, including AI spirituality, community events, and updates on moderation practices.
This open thread celebrates the winners of the ACX/Metaculus Forecasting Contest and encourages community interaction on various topics.
2.6Open Thread 423
2026-03-02 • finance open thread announcements ai
This weekly open thread invites discussions while announcing Stephen Grugett's MNX project and clarifying the author's previous tone on a related article.