Elicit is an AI research assistant with 2M+ users that searches 138M papers and automates literature reviews, founded by AI alignment researchers from Ought and funded by Coefficient Giving (formerly Open Philanthropy, $31M total). The platform achieved 90%+ extraction accuracy and claims 80% time savings for systematic reviews, though independent validation of these claims is limited.
Constitutional AIApproachConstitutional AIConstitutional AI is Anthropic's methodology using explicit principles and AI-generated feedback (RLAIF) to train safer models, achieving 3-10x improvements in harmlessness while maintaining helpfu...Quality: 70/100Refusal TrainingApproachRefusal TrainingRefusal training achieves 99%+ refusal rates on explicit harmful requests but faces 1.5-6.5% jailbreak success rates (UK AISI 2025) and 12-43% over-refusal on legitimate queries. While necessary fo...Quality: 63/100AI-Augmented ForecastingApproachAI-Augmented ForecastingAI-augmented forecasting combines AI computational strengths with human judgment, achieving 5-15% Brier score improvements and 50-200x cost reductions compared to human-only forecasting. However, A...Quality: 54/100
Analysis
Model Organisms of MisalignmentAnalysisModel Organisms of MisalignmentModel organisms of misalignment is a research agenda creating controlled AI systems exhibiting specific alignment failures as testbeds. Recent work achieves 99% coherence with 40% misalignment rate...Quality: 65/100RoastMyPostProjectRoastMyPostRoastMyPost is an LLM tool (Claude Sonnet 4.5 + Perplexity) that evaluates written content through multiple specialized AI agents—fact-checking, logical fallacy detection, math verification, and mo...Quality: 35/100
Risks
Reward HackingRiskReward HackingComprehensive analysis showing reward hacking occurs in 1-2% of OpenAI o3 task attempts, with 43x higher rates when scoring functions are visible. Mathematical proof establishes it's inevitable for...Quality: 91/100
Other
Scalable OversightResearch AreaScalable OversightProcess supervision achieves 78.2% accuracy on MATH benchmarks (vs 72.4% outcome-based) and is deployed in OpenAI's o1 models, while debate shows 60-80% accuracy on factual questions with +4% impro...Quality: 68/100RLHFResearch AreaRLHFRLHF/Constitutional AI achieves 82-85% preference improvements and 40.8% adversarial attack reduction for current systems, but faces fundamental scalability limits: weak-to-strong supervision shows...Quality: 63/100Eli LiflandPersonEli LiflandBiographical profile of Eli Lifland, a top-ranked forecaster and AI safety researcher who co-authored the AI 2027 scenario forecast and co-founded the AI Futures Project. The page documents his for...Quality: 58/100Andreas StuhlmüllerPersonAndreas StuhlmüllerCo-founder of Elicit (formerly Ought), an AI research tool focused on language model assistance for research tasks. Pioneer in using ML to assist human reasoning and decision-making.
Concepts
Large Language ModelsCapabilityLarge Language ModelsComprehensive analysis of LLM capabilities showing rapid progress from GPT-2 (1.5B parameters, 2019) to GPT-5 and Gemini 2.5 (2025), with training costs growing 2.4x annually and projected to excee...Quality: 60/100Wikipedia and AI ContentConceptWikipedia and AI ContentWikipedia faces three-way AI pressure: being consumed as training data (47.9% of ChatGPT citations), infiltrated by AI content (~5% of new articles), and losing traffic to AI summaries (8% decline ...Quality: 56/100AI-Assisted Knowledge ManagementConceptAI-Assisted Knowledge ManagementA comprehensive survey of AI-assisted knowledge management tools (Obsidian plugins, Notion AI, NotebookLM, RAG frameworks) with specific cost figures ($4-6/page for Longterm Wiki pipeline) and a co...Quality: 48/100Agentic AICapabilityAgentic AIAnalysis of agentic AI capabilities and deployment challenges, documenting industry forecasts (40% of enterprise apps by 2026, $199B market by 2034) alongside implementation difficulties (40%+ proj...Quality: 68/100
Organizations
SamotsvetyOrganizationSamotsvetyElite forecasting group Samotsvety dominated INFER competitions 2020-2022 with relative Brier scores twice as good as competitors, providing influential probabilistic forecasts including 28% TAI by...Quality: 61/100Coefficient GivingOrganizationCoefficient GivingCoefficient Giving (formerly Open Philanthropy) has directed $4B+ in grants since 2014, including $336M to AI safety (~60% of external funding). The organization spent ~$50M on AI safety in 2024, w...Quality: 55/100AI Futures ProjectOrganizationAI Futures ProjectAI Futures Project is a nonprofit co-founded in 2024 by Daniel Kokotajlo, Eli Lifland, and Thomas Larsen that produces detailed AI capability forecasts, most notably the AI 2027 scenario depicting ...Quality: 50/100LessWrongOrganizationLessWrongLessWrong is a rationality-focused community blog founded in 2009 that has influenced AI safety discourse, receiving $5M+ in funding and serving as the origin point for ~31% of EA survey respondent...Quality: 44/100GoodfireOrganizationGoodfireGoodfire is a well-funded AI interpretability startup valued at $1.25B (Feb 2026) developing mechanistic interpretability tools like Ember API to make neural networks more transparent and steerable...Quality: 68/100FutureSearchOrganizationFutureSearchFutureSearch is an AI forecasting startup founded by former Metaculus leaders that combines LLM research agents with human judgment, demonstrating some prediction accuracy but facing uncertain comm...Quality: 50/100