Currently Reading

• Good Strategy, Bad Strategy: The Difference and Why it Matters (Richard Rumelt, 2011)

2026

• If Anyone Builds It Everyone Dies: Why Superhuman AI Would Kill Us All (Eliezer Yudkowsky and Nate Soares, 2026)
• A Christmas Carol (Charles Dickens, 1843)

2025

• BlueDot Biosecurity Course (BlueDot Impact, 2025)
• Why I Write: Essay Collection (George Orwell, 1946)
• The Time Machine (H. G. Wells, 1895)
• Brave New World (Aldous Huxley, 1932)
• The Coaching Habit: Say Less, Ask More & Change The Way You Lead Forever (Michael Bungay Stanier, 2016)
• Animal Farm (George Orwell, 1945)
• The Metamorphosis (Franz Kafka, 1915)
• The Little Prince (Antoine de Saint-Exupéry, 1943)
• 1984 (George Orwell, 1949)
• How to Actually Change Your Mind (Eliezer Yudkowsky, 2009)
• Inadequate Equilibria: How and Why Civilizations Get Stuck (Eliezer Yudkowsky, 2017)
• Map and Territory (Eliezer Yudkowsky, 2009)
• Digital Minimalism: Choosing a Focused Life in a Noisy World (Cal Newport, 2019)
• The Big Score: The Billion Dollar Story of Silicon Valley (Michael S Malone, 1985)
• Quran 30 for 30: Thematic Tasfir (Omar Suleiman & Ismail Kamdar, 2025)
• From Counterculture to Cyberculture: Stewart Brand, the Whole Earth Network, and the Rise of Digital Utopianism (Fred Turner, 2008)
• Slouching Towards Bethlehem (Joan Didion, 1968)
• To the Finland Station: A Study in the Writing and Acting of History (Edmund Wilson, 1940)
• Bury the Chains: The British Struggle to Abolish Slavery (Adam Hochschild, 2005)
• How the World Became Rich: The Historical Origins of Economic Growth (Mark Koyama & Jared Rubin, 2022)
• The Jewel House: Elizabethan London and the Scientific Revolution (Deborah Harkness, 2007)
• The Final Pagan Generation (Edward J. Watts, 2015)
• The Rise of Christianity (Rodney Stark, 1996)
• The Kingdom (Emmanuel Carrère, 2014)
• The Structure of Scientific Revolutions (Thomas Kuhn, 1962)

2024

• Zero to One: Notes on Startups, or How to Build the Future (Peter Thiel & Blake Masters, 2014) [my summary here]
• Harry Potter and the Methods of Rationality (Eliezer Yudkowsky, 2015)

2022

• Superintelligence: Paths, Dangers, Strategies (Nick Bostrom, 2014)
• Algorithms to Live By: The Computer Science of Decisions (Brian Christian & Tom Griffiths, 2016)
• The Scout Mindset: Why Some People See Things Clearly and Others Don't (Julia Galef, 2021)
• Superforecasting: The Art of Science and Prediction (Philip Tetlock & Dan Gardner, 2015)
• Human Compatible: Artificial Intelligence and the Problem of Control (Stuart Russell, 2019)
• Doing Good Better: Effective Altruism and How You Can Make a Difference (William MacAskill, 2015)

2021

• The Precipice: Existential Risk and the Future of Humanity (Toby Ord, 2020)
• Why We Sleep: The New Science of Sleep and Dreams (Matthew Walker, 2017)
• How to Win Friends and Influence People (Dale Carnegie, 1936)

March 2026

• Half-assing it with everything you've got (mindingourway.com)
• AI for AI safety (joecarlsmith.substack.com)
• Paths and waystations in AI safety (joecarlsmith.substack.com)
• When should we worry about AI power-seeking? (joecarlsmith.substack.com)
• I underestimated AI capabilities (again) - by Ajeya Cotra (planned-obsolescence.org)
• Replacing guilt (mindingourway.com)
• You're not burning out because you're tired — EA Forum (forum.effectivealtruism.org)
• A personal take on why (and why not) to work on AI safety at Open Philanthropy (catherinebrewer.substack.com)
• 🟡 US and Israel strike Iran, Iran retaliates against US allies, Anthropic designated a supply chain risk || Global Risks Weekly Roundup #9/2026 (blog.sentinel-team.org)

February 2026

• Here's to the Polypropylene Makers — LessWrong (lesswrong.com)
• What is it to solve the alignment problem? - Joe Carlsmith (joecarlsmith.com)
• How do we solve the alignment problem? - Joe Carlsmith (joecarlsmith.com)
• Statement from Dario Amodei on our discussions with the Department of War \ Anthropic (anthropic.com)
• We hosted a dessert invention contest - by Ben James (benbyfax.substack.com)
• The Persona Selection Model: Why AI Assistants might Behave like Humans (alignment.anthropic.com)
• The ML ontology and the alignment ontology — LessWrong (lesswrong.com)
• 🟩 US buildup in Middle East continues, Russian vessel en route to Cuba, AI time horizons lengthen | Global Risks Weekly Roundup #8/2026 (blog.sentinel-team.org)
• We're in Triage Mode for AI Policy - by Miles Brundage (milesbrundage.substack.com)
• The stakes of AI moral status - Joe Carlsmith (joecarlsmith.com)
• Three positive updates I made about technical grantmaking at Coefficient Giving (fka Open Phil) — LessWrong (lesswrong.com)
• Changing the world for the worse — LessWrong (lesswrong.com)
• Animal welfare in 1800 — EA Forum (forum.effectivealtruism.org)
• The Unilateralist’s Curse and the Case for a Principle of Conformity (pmc.ncbi.nlm.nih.gov)
• [REPOST] Epistemic Learned Helplessness (slatestarcodex.com)
• Against neutrality about creating happy lives - Joe Carlsmith (joecarlsmith.com)
• Actually possible: thoughts on Utopia - Joe Carlsmith (joecarlsmith.com)
• Request for Proposals: Improving Capability Evaluations (coefficientgiving.org)
• [Closed] Request for Proposals: Studying and Forecasting the Real-World Impacts of Systems Built from LLMs (coefficientgiving.org)
• [Closed] Request for Proposals: Benchmarking LLM Agents on Consequential Real-World Tasks (coefficientgiving.org)
• Research directions Open Phil wants to fund in technical AI safety — LessWrong (lesswrong.com)
• Open Philanthropy Technical AI Safety RFP - $40M Available Across 21 Research Areas — LessWrong (lesswrong.com)
• New roles on my team: come build Open Phil's technical AI safety program with me! — EA Forum (forum.effectivealtruism.org)
• Comments on OpenPhil's Interpretability RFP — LessWrong (lesswrong.com)
• On future people, looking back at 21st century longtermism - Joe Carlsmith (joecarlsmith.com)
• Killing the ants - Joe Carlsmith (joecarlsmith.com)
• Thoughts on being mortal - Joe Carlsmith (joecarlsmith.com)
• On clinging - Joe Carlsmith's Substack (joecarlsmith.substack.com)
• The costs of caution (planned-obsolescence.org)
• Leaving Open Philanthropy, going to Anthropic - Joe Carlsmith (joecarlsmith.com)
• By 2050 we could get "10,000 years of technological progress" (youtube.com)
• Functional Representations of Emotion in a Large Language Model (transformer-circuits.pub)
• Why we should expect ruthless sociopath ASI — LessWrong (lesswrong.com)
• The attempted kidnapping at my house - by Aella (aella.substack.com)
• A multi-level postmortem of how our whole house got badly poisoned — LessWrong (lesswrong.com)
• Church Planting: When Venture Capital Finds Jesus — LessWrong (lesswrong.com)
• 🟩 US sends second aircraft carrier to Middle East, European nuclear deterrent discussed, pandemic preparedness deprioritized at NIAID | Global Risks Weekly Roundup #7/2026 (blog.sentinel-team.org)
• You Have Only X Years To Escape Permanent Moon Ownership (astralcodexten.com)
• Aligning to Virtues — LessWrong (lesswrong.com)
• Life at the Frontlines of Demographic Collapse — LessWrong (lesswrong.com)
• Do Less. - by Cate Hall - Useful Fictions (usefulfictions.substack.com)
• The Missing Genre: Heroic Parenthood - You can have kids and still punch the sun — LessWrong (lesswrong.com)
• Toys with the highest play-time and lowest clean-up-time (joannabregan.substack.com)
• Ending Parking Space Saving — LessWrong (lesswrong.com)
• after my dad died, we found the love letters (jenn.site)
• my dad could still be alive, but he's not. (jenn.site)
• Making every researcher seek grants is a broken model — LessWrong (lesswrong.com)
• 🟩 Anthropic and OpenAI release new models, US-Iran talks continue, illegal biolab in the US, nuclear arms treaty expires | Global Risks Weekly Roundup #6/2026 (blog.sentinel-team.org)
• Angels-on-the-Shoulder: 5 AI Tools for Better Decisions (forethought.org)
• I miss thinking hard. (jernesto.com)
• AlgZoo: uninterpreted models with fewer than 1,500 parameters (alignment.org)
• Eli's shortform feed — LessWrong (lesswrong.com)
• Jeremy Bentham - Wikipedia (en.wikipedia.org)
• Different senses in which two AIs can be “the same” — LessWrong (lesswrong.com)
• the case for CoT unfaithfulness is overstated — LessWrong (lesswrong.com)
• Extending control evaluations to non-scheming threats (blog.redwoodresearch.org)
• It Is Reasonable To Research How To Use Model Internals In Training — LessWrong (lesswrong.com)
• 50 things I know (usefulfictions.substack.com)
• Berkson's paradox - Wikipedia (en.wikipedia.org)
• Burnout is breaking a sacred pact (usefulfictions.substack.com)
• bye, mom (aella.substack.com)
• The seam through the center of things (usefulfictions.substack.com)
• The lies I used to tell myself (usefulfictions.substack.com)
• In praise of quitting (usefulfictions.substack.com)
• Aliveness and where to find it - by Ben James - Ben by Fax (substack.com)
• Are you stuck in movie logic? (usefulfictions.substack.com)
• Rightness is a prison (usefulfictions.substack.com)
• How to increase your surface area for luck (usefulfictions.substack.com)
• Tying yourself to the mast (usefulfictions.substack.com)
• Everyone’s the hero of their own story (usefulfictions.substack.com)
• How to instantly be better at things (usefulfictions.substack.com)
• Fuck willpower (usefulfictions.substack.com)
• Learn to love the Moat of Low Status (usefulfictions.substack.com)
• Are you a jerk, or a liar? (usefulfictions.substack.com)
• Everyone here is in a cult (usefulfictions.substack.com)
• Crossing the cringe minefield (usefulfictions.substack.com)
• Barnum effect - Wikipedia (en.wikipedia.org)
• 37 things people love about Sasha Chapin (usefulfictions.substack.com)
• I used to be a high-performing robot (usefulfictions.substack.com)
• There are nine wolves inside of you (usefulfictions.substack.com)
• Maybe you’re not Actually Trying (usefulfictions.substack.com)
• Paper Highlights of January 2026 (aisafetyfrontier.substack.com)
• How Nature Became a 'Prestige' Journal (asimov.press)
• TPU Deep Dive (henryhmko.github.io)
• Opinionated Takes on Meetups Organizing — LessWrong (lesswrong.com)
• Why read novels? (dynomight.net)
• On neural scaling and the quanta hypothesis (ericjmichaud.com)
• “Features” aren’t always the true computational primitives of a model, but that might be fine anyways — LessWrong (lesswrong.com)
• The Hidden Cost of Our Lies to AI — LessWrong (lesswrong.com)
• A playbook for field strategy (blog.bluedot.org)
• Aliveness and where to find it (benbyfax.substack.com)
• Fake thinking and real thinking (joecarlsmith.substack.com)

January 2026

• How AI Is Learning to Think in Secret — LessWrong (lesswrong.com)
• Ilya Sutskever – We're moving from the age of scaling to the age of research - YouTube (youtube.com)
• Can we safely automate alignment research? — LessWrong (lesswrong.com)
• Reflections on 2025 (samuelalbanie.substack.com)
• The Assistant Axis: Situating and Stabilizing the Default Persona... (arxiv.org)
• We Let an AI Talk To Another AI. Things Got Really Weird. | Kyle Fish, Anthropic - YouTube (youtube.com)
• Alignment remains a hard, unsolved problem — LessWrong (lesswrong.com)
• On Agency (sebastianfarquhar.com)
• Why Every Scientist That Wanted Mirror Bacteria Changed Their Mind - YouTube (youtube.com)
• Film Study for Research (jsteinhardt.stat.berkeley.edu)
• The Most Important Graph in AI Right Now | Beth Barnes, CEO of METR - YouTube (youtube.com)
• Paper Highlights of December 2025 - by Johannes Gasteiger (aisafetyfrontier.substack.com)
• Book Review: Seeing Like A State (slatestarcodex.com)
• Artificial meat is harder than artificial intelligence — Lewis Bollard - YouTube (youtube.com)
• AI-engineered diseases are coming. Here's the plan to stop them. | Andrew Snyder-Beattie - YouTube (youtube.com)
• Oversight Assistants: Turning Compute into Understanding — LessWrong (lesswrong.com)
• The Redwood Research podcast - YouTube (youtube.com)
• 🟩 US attacks Venezuela and captures Maduro, threatens Iran, Mexico, Colombia and Denmark || Global Risks Weekly Roundup #1/2026 (blog.sentinel-team.org)
• Recent LLMs can use filler tokens or problem repeats to improve (no-CoT) math performance — LessWrong (lesswrong.com)
• Recent LLMs can do 2-hop and 3-hop latent (no-CoT) reasoning on natural facts — LessWrong (lesswrong.com)
• Unsupervised decoding of encoded reasoning using language model... (arxiv.org)

December 2025

• Turning 20 in the probable pre-apocalypse — LessWrong (lesswrong.com)
• Why Not Just Train For Interpretability? — LessWrong (lesswrong.com)
• How to be more agentic (usefulfictions.substack.com)
• Mini Blog Post 3: Become a person who Actually Does Things — Neel Nanda (neelnanda.io)
• Emergent Introspective Awareness in Large Language Models (transformer-circuits.pub)
• The UK Recovery Trial for Covid-19 treatments helped save a million l… (archive.ph)
• From Warp Speed to 100 Days—Asterisk (asteriskmag.com)
• The behavioral selection model for predicting AI motivations — LessWrong (lesswrong.com)
• So you’ve taken over the world - by Owen Cotton-Barratt (strangecities.substack.com)