rational
Issue #217: Full List
19 June, 2022 // View curated list
# Instrumental
Perils of optimizing in social contexts // owencb, 2 min
Don't Over-Optimize Things // owencb, 5 min
Favourite new AI productivity tools? // gabe-mukobi, 1 min
The best 'free solo' (rock climbing) video // Kenny, 3 min
Things That Make Me Enjoy Giving Career Advice // neel-nanda-1, 10 min
What's the goal in life? // k8n, 1 min
# Epistemic
Our mental building blocks are more different than I thought // marius-hobbhahn, 17 min
[Link] "The madness of reduced medical diagnostics" by Dynomight // Kenny, 1 min
[Link] New SEP article on Bayesian Epistemology // alenglander, 1 min
# Ai
Contra Hofstadter on GPT-3 Nonsense // rictic, 1 min
Humans are very reliable agents // alyssavance, 3 min
Slow motion videos as AI risk intuition pumps // Andrew_Critch, 1 min
Contra EY: Can AGI destroy us without trial & error? // nikita-sokolsky, 18 min
A central AI alignment problem: capabilities generalization, and the sharp left turn // So8res, 12 min
Yes, AI research will be substantially curtailed if a lab causes a major disaster // lc, 2 min
Why all the fuss about recursive self-improvement? // So8res, 8 min
On A List of Lethalities // Zvi, 66 min
Scott Aaronson is joining OpenAI to work on AI safety // peterbarnett, 1 min
Pivotal outcomes and pivotal processes // Andrew_Critch, 5 min
A transparency and interpretability tech tree // evhub, 23 min
wrapper-minds are the enemy // nostalgebraist, 9 min
A claim that Google's LaMDA is sentient // ben-livengood, 1 min
Resources I send to AI researchers about AI safety // Vael Gates, 10 min
Ten experiments in modularity, which we'd like you to run! // TheMcDouglas, 11 min
FYI: I’m working on a book about the threat of AGI/ASI for a general audience. I hope it will be of value to the cause and the community // Darren McKee, 2 min
Why so little AI risk on rationalist-adjacent blogs? // grant-demaree, 8 min
"Brain enthusiasts" in AI Safety // jan-2, 11 min
Blake Richards on Why he is Skeptical of Existential Risk from AI // mtrazzi, 5 min
What's the "This AI is of moral concern." fire alarm? // quintin-pope, 2 min
A Modest Pivotal Act // anonymousaisafety, 5 min
Alignment Risk Doesn't Require Superintelligence // JustisMills, 2 min
Continuity Assumptions // Jan_Kulveit, 5 min
Let's not name specific AI labs in an adversarial context // acylhalide, 1 min
How much does cybersecurity reduce AI risk? // Darmani, 1 min
Vael Gates: Risks from Advanced AI (June 2022) // Vael Gates, 15 min
AI Risk, as Seen on Snapchat // dkirmani, 1 min
What are all the AI Alignment and AI Safety Communication Hubs? // Gunnar_Zarncke, 1 min
Investigating causal understanding in LLMs // marius-hobbhahn, 15 min
What if LaMDA is indeed sentient / self-aware / worth having rights? // RomanS, 1 min
Can DALL-E understand simple geometry? // KingSupernova, 1 min
Alignment research for "meta" purposes // acylhalide, 1 min
Quantifying General Intelligence // JasonBrown, 15 min
How Do You Quantify [Physics Interfacing] Real World Capabilities? // DragonGod, 4 min
Will vague "AI sentience" concerns do more for AI safety than anything else we might do? // alenglander, 1 min
OpenAI: GPT-based LLMs show ability to discriminate between its own wrong answers, but inability to explain how/why it makes that discrimination, even as model scales // aditya-jain, 1 min
Grokking “Semi-informative priors over AI timelines” // anson.ho, 16 min
A possible AI-inoculation due to early "robot uprising" // shminux, 1 min
Value extrapolation vs Wireheading // Stuart_Armstrong, 1 min
Crystalizing an agent's objective: how inner-misalignment could work in our favor // joshua-clymer, 5 min
The beautiful magical enchanted golden Dall-e Mini is underrated // p.b., 1 min
Training Trace Priors // adam-jermyn, 5 min
High Powers Over Physics // DragonGod, 4 min
Agent level parallelism // johannes-c-mayer, 1 min
Has there been any work on attempting to use Pascal's Mugging to make an AGI behave? // Chris_Leong, 1 min
Why multi-agent safety is important // akbir-khan, 12 min
What are some smaller-but-concrete challenges related to AI safety that are impacting people today? // nonzerosum, 1 min
How much stupider than humans can AI be and still kill us all through sheer numbers and resource access? // shminux, 1 min
Is there an unified way to make sense of ai failure modes? // walking_mushroom, 1 min
High Powers of Physics Manipulation // DragonGod, 4 min
Multigate Priors // adam-jermyn, 3 min
Can you MRI a deep learning model? // yair-halberstadt, 1 min
Breaking Down Goal-Directed Behaviour // Oliver Sourbut, 2 min
Cryptographic Life: How to transcend in a sub-lightspeed world via Homomorphic encryption // Golol, 3 min
# Meta-ethics
Lifeguards // akash-wasil, 3 min
Pure Altruism // Matt Goldwater, 4 min
# Anthropic
Why don't we think we're in the simplest universe with intelligent life? // ADifferentAnonymous, 1 min
UFOs as USA PsychOp // Robin Hanson, 4 min
# Decision theory
Georgism, in theory // Stuart_Armstrong, 5 min
Was the Industrial Revolution The Industrial Revolution? // Davis Kedrosky, 14 min
Is there a worked example of Georgian taxes? // Dagon, 1 min
An Approach to Land Value Taxation // harsimony, 5 min
1689: Uncovering the World New Institutionalism Created // Davis Kedrosky, 11 min
# Math and cs
A Butterfly's View of Probability // gabriel-wu, 14 min
Crypto-fed Computation // aaguirre, 9 min
Do yourself a FAVAR: security mindset // lcmgcd, 2 min
Intuitive Explanation of AIXI // thomas-larsen, 6 min
Optimization power as divergence from default trajectories // joshua-clymer, 6 min
#SAT with Tensor Networks // adam-jermyn, 2 min
Why the Kaldor-Hicks criterion can be non-transitive // Rupert, 1 min
# Books
Forecasting Fusion Power // daniel-kokotajlo, 1 min
Your Book Review: The Future Of Fusion Energy // Scott Alexander, 22 min
# Community
I applied for a MIRI job in 2020. Here's what happened next. // ViktoriaMalyasova, 9 min
Announcing the Clearer Thinking Regrants program // spencerg, 1 min
BBC Future covers progress studies // jasoncrawford, 3 min
# Culture war
Against Active Shooter Drills // Zvi, 8 min
What's the information value of government hearings? // Kenny, 2 min
Why did Russia invade Ukraine? // Bohaska
# Misc
In defense of flailing, with foreword by Bill Burr // lc, 4 min
Infohazards and inferential distances // acylhalide, 6 min
Security analysis of 'cloud chemistry labs'? // Kenny, 1 min
Peer Review: Nightmares // Scott Alexander, 8 min
Why Be Romantic // Robin Hanson, 3 min
# Podcasts
Who Are You EP 01: Seth Jordan on Social Threefolding // The Jim Rutt Show, 79 min
#132 - Nova DasSarma on why information security may be critical to the safe development of AI systems // , 162 min
164 – Pregnancy is Body Horror // The Bayesian Conspiracy, 112 min
Currents 064: Michael Garfield and J.F. Martel on Art x AI // The Jim Rutt Show, 61 min
# Videos of the week
Large Language Models as a Cultural Technology // Simons Institute, 15 min