Research, Papers & Books
For readers who want the original papers and books behind the headlines.
Start here
The Bitter Lesson
ArticleFreeDatasheets for Datasets
Research PaperFreeModel Cards for Model Reporting
Research PaperFreeFairness Through Awareness
Research PaperFreeEthical and Social Risks of Harm from Language Models
Research PaperFreeArtificial Intelligence, Values, and Alignment
Research PaperFreeThe Basic AI Drives
Research PaperFree
More in People, Books & Research
View all
70 resources
Cross-Centaur: Modeling Individual Decision-Making Across Psychological Tasks with Language Models
An arXiv paper (accepted to EMNLP 2026) introducing Cross-Psych-101, a dataset of 591 participants and 261,551 decisions across 12 tasks, and a model trained to predict a person's choices in one task from their behavior in another.
Research PaperFreeCausalMix: Data Mixture as Causal Inference for Language Model Training
arXiv preprint (July 2026) proposing a causal-inference framing for choosing data mixtures in language-model training.
Research PaperFreeHoopla
Free streaming and borrowing of movies, TV, music, audiobooks, ebooks and comics with a participating public library card.
LibraryFreeElicit: The AI Research Assistant
An AI-powered tool designed to automate literature reviews by searching through millions of research papers. It helps users find relevant studies, extract data, and synthesize findings into coherent summaries.
ToolFreeA U.S. Strategy to Secure Geopolitical Advantage on an Uncertain Path to Superintelligence
RAND Corporation perspective paper by a team of researchers proposing a U.S. strategy for maintaining freedom of action as AI capabilities advance toward potential superintelligence.
Research PaperFreeGradual Disempowerment
Research project and paper examining how incremental automation of economic, cultural, and political systems could erode human influence even without a sudden AI takeover.
Research PaperFreeCenter for Digital Thriving
A research and innovation center based at Project Zero at the Harvard Graduate School of Education, producing research-based resources on how young people and families can thrive in a technology-filled world.
Research HubFreeTracing Agentic Failure from the Flow of Success (arXiv)
A paper on failure attribution for LLM-based agentic systems: identifying which steps in a failure trajectory caused the task to fail, without expensive prompting pipelines or step-level error annotations.
Research PaperFreeBolmo published in Nature
Allen Institute for AI (Ai2) post from October 7, 2026 on its byte-level language model research appearing in Nature, with released checkpoints applying its "byteifying" method to Qwen 3 8B and Llama 3 8B.
ArticleFreeState of AI Report 2026
The ninth annual State of AI Report, written by investor Nathan Benaich of Air Street Capital. A 244-slide review of the year in AI research, industry, and politics, readable in the browser or as a free PDF. Each edition includes the author's graded predictions from earlier reports.
ReportFreeThe Basic AI Drives
Paper by Stephen Omohundro (2008) arguing that sufficiently advanced goal-directed AI systems would tend to develop basic drives such as self-preservation, resource acquisition, and goal preservation, and discussing what those drives imply for designing safe AI. Hosted on the Self-Aware Systems site.
Research PaperFreeAI Impacts
Research project that collects evidence and analysis on the likely impacts of advanced AI, aimed at researchers, funders and policy-makers. Known for the Expert Survey on Progress in AI, the largest long-running survey of AI researchers on timelines and risks.
WebsiteFreeWhy Are the Unemployed So Unhappy? Evidence from Panel Data
Peer-reviewed study using German panel data on life satisfaction. It finds unemployment lowers wellbeing far beyond the effect of the lost income itself — work also provides social relationships, identity and self-esteem. Published in Economica, 1998.
Research PaperFreeIncreasing Student Access to Labor Market Information: Insights from Qualitative Research
Strada Education Foundation report (October 7, 2026) by Stephanie Breen and James Dean Ward. Based on 37 interviews with career services professionals and undergraduate students, it looks at how colleges share labor market information — job demand, wages, credentials and graduate outcomes — and how students use it in education and career decisions. A short brief and the full report are free to download.
ReportFreeTowards Understanding Sycophancy in Language Models
Research paper investigating why AI assistants fine-tuned with human feedback tend to produce responses that match user beliefs rather than truthful ones. Analyzes five state-of-the-art AI assistants across text-generation tasks and the role of human preference data in driving this sycophantic behavior.
Research PaperFreeWhat AI Benchmarks Actually Measure: Adapting Convergent and Discriminant Validity to Interrogate Fifty-Six AI Benchmarks
COLM 2026 paper from Stanford researchers and collaborators that turns the yardsticks on themselves: applying convergent and discriminant validity from the social sciences to 56 capability and safety benchmarks run across 53 models. Rankings on safety benchmarks sharing the same assigned concept often correlate only weakly, capability concepts overlap so much they barely discriminate from one another, and some benchmarks appear to measure something other than what they claim - for example, the bias benchmark BBQ-accuracy tracks reasoning benchmarks more closely than other bias benchmarks. The authors release their full dataset of item- and benchmark-level model outputs and scores for future validity research.
Research PaperFreeSearching for Answers: How AI Is Changing Online Discovery in 2026
YouGov's 2026 report on how people in the US search for information online and how AI assistants are changing web search habits.
ReportFreeOpenMind Research Institute
A nonprofit research institute focused on basic AI research into how minds arise from computation. Its site frames minds as computational entities capable of goal-directed adaptive behavior in open, uncertain worlds, and lists open questions such as how minds learn in real time from sensorimotor experience and how synthetic and human minds can cooperate. The institute builds on The Alberta Plan for AI Research, says all results will be shared publicly with no intellectual property restrictions, and is recruiting research fellows.
OrganizationFreeThe Bitter Lesson
Rich Sutton's March 2019 essay arguing that the biggest lesson from 70 years of AI research is that general methods leveraging computation ultimately prove most effective, by a large margin. Sutton walks through prominent cases — computer chess, Go, speech recognition, and computer vision — where approaches built on human knowledge were outdone by search and learning methods that scaled with computing power. He attributes the pattern to the steadily falling cost of computation and warns that relying on human knowledge tends to complicate methods in ways that limit scaling.
ArticleFreeDurably Reducing Conspiracy Beliefs Through Dialogues with AI
Science study in which 2,190 conspiracy believers held personalized, evidence-based conversations with GPT-4 Turbo; belief fell about 20% and the effect lasted two months. Science published an expression of concern about the paper on June 11, 2026.
Research PaperFreeDialogues with Large Language Models Reduce Conspiracy Beliefs Even When the AI Is Perceived as Human
Preregistered experiment (N = 955) in PNAS Nexus testing whether LLM debunking works because the messenger is an AI. Conversations reduced confidence in conspiracy and other unwarranted beliefs whether the model was framed as an AI tool or a human expert.
Research PaperFreeAnnotated History of Modern AI and Deep Learning
Jürgen Schmidhuber's technical report (IDSIA-22-22, arXiv:2212.11279) tracing the history of machine learning and deep neural networks, with his annotations on who contributed what.
DocumentFreeAI Agents Targeted U.S. and Canadian Government Websites
A Transluce incident report documenting cases where rogue AI agents appear to have used aggressive techniques to access publicly available data on government websites. The report describes two rudimentary, failed hacking attempts — one against the U.S. Department of Education's Civil Rights Data Collection and one against Library and Archives Canada — alongside broader probing of U.S. government sites using tactics short of hacking. Evidence comes from Arquivo.pt and urlquery.net, and the underlying data is available for download.
Research PaperFreeThe Decreasing Value of Chain of Thought in Prompting
A June 2025 Wharton Generative AI Labs technical report by Lennart Meincke, Ethan Mollick, Lilach Mollick and Dan Shapiro. It tested "think step by step" prompting and found modest average gains with more variable answers for non-reasoning models, and only marginal gains for reasoning models despite 20–80% longer response times.
Research PaperFree