Analysis — page 2
Cheshire Academy uses a traffic-light system to govern AI use in class
The 400-student Connecticut boarding school labels every assignment green, yellow, or red — and lets teachers pick the tools.
METR finds AI is accelerating cyber and math research — but not AI itself
A METR study documents lumpy AI-driven progress across fields, while SPADE and Hawkeye show how models are starting to bootstrap their own training.
Children learn language on 100M words. LLMs need 100,000x more
The data efficiency gap between toddlers and frontier models is now the defining puzzle for both cognitive science and AI research.
AI executives misread the public backlash as a messaging problem
More than half of Americans under 30 are more worried than excited about AI, a 24-point swing in five years. The industry's diagnosis keeps missing.
Pew finds 35% of post-ChatGPT web pages show signs of AI authorship
A near-half-million-page Pew analysis says .com sites are 10x more likely to be AI-written than .edu or .gov domains.
Rumman Chowdhury: AI consciousness debate is a liability shield
Op-ed argues frontier labs and effective altruists are converging on a framing that shields AI companies from product-liability suits.
Vivodyne bets AI can't cure cancer without human tissue data
The Khosla-backed startup opened a robotic lab growing 20 human tissue types to feed AI models the causal biology they lack.
Stanford's AI Observatory finds Anthropic filters out 48% of Claude conversations
A new independent dataset shows sensitive AI use — companionship, health, harassment — runs far higher than company reports admit.
Anthropic's Dario Amodei calls AI backlash 'a crisis of trust'
Amodei pushes back on investor Gavin Baker's claim that his AI warnings have fueled US opposition to data centers and new regulation.
Tim O'Reilly says frontier AI labs are reading the future wrong
The publisher argues open-source AI, not bigger models from OpenAI or Anthropic, will define the next wave — even as AI hollows out his book business.
US teens split on AI: 57% use chatbots to search, only 12% for emotional support
A Pew survey and student interviews show teenagers using AI heavily for schoolwork and search while pushing back on creative and emotional uses.
Hinton, Li, and Ng defend open AI at Ai4, split on how open
Three of AI's most cited researchers agreed openness matters. They disagreed on open weights, China, and where regulation belongs.
AI model maps 766 schizophrenia genes, 641 of them new, in Nature Genetics study
Researchers analyzed data from 102,000 people and six brain regions to reveal a coordinated genetic network behind the disorder.
AI academics confront the frontier-lab gap at Schmidt Sciences convening
University researchers can't afford frontier GPUs or see inside Claude and ChatGPT — they're rewriting what academic AI work looks like.
Zuckerberg publishes 6,500-word AI manifesto pitching open superintelligence
Meta's CEO lays out an infrastructure, distribution, and regulatory plan built around free access and open-weight models.
AI slop backlash forces Meta, Google to pull generative features
Meta killed its Instagram deepfake tool after three days of public outcry, as nearly half of young Americans say generative AI does more harm than good.
Peer review buckles as paper volume climbs 5.6% a year and AI floods journals
Editors now email 30 reviewers to land one, and AI researchers are quietly abandoning journals for blogs.
Eric Schmidt argues AI agents, not AlphaFold clones, will accelerate science
The former Google CEO says the Protein Data Bank took 53 years and $21B to build — and most sciences will never get that lucky.
Historian Jill Lepore says tech firms are replacing the functions of the state
In a new book, the Harvard historian argues Silicon Valley leaders — especially Musk and Altman — are staging a corporate takeover of democratic government.
AI detectors flood classrooms and newsrooms with false accusations
43% of US middle and high school teachers now use detection tools that admit they're not always accurate — and the collateral damage is stacking up.
ChatTJB, a chatbot that is just one guy, logs 30,000 queries
Tucker Bryant's ChatTJB answers prompts by hand — no LLM involved. A $6,000 San Francisco billboard pushed it past 5,000 prompts an hour at peak.
Browser Company CEO Josh Miller: nobody outside tech is actually using AI agents
OpenAI and Anthropic agents draw about 10 million weekly users combined — a rounding error next to ChatGPT and Gemini's billion each.
Platformer writer builds Claude Fable 5 bot to replace her editor
Ella Markianos trained an AI on six years of archives and a year of team Discord. It landed 30% useful edits and one factual error every two columns.
Hank Green calls his AI use 'not healthy' as chatbot dependency grows
The YouTuber's step-back highlights a middle zone of compulsive LLM use that sits between benign help and full psychiatric crisis.
Stay ahead of everyone in AI.
The tightly edited AI news email engineers, founders, and investors actually open. One email. Every weekday. Five minutes to finish.
The briefing read inside teams at