rationalnewsletter

Issue #154

Future AI systems will learn as people do — without relying on labeled data sets. Facebook developed SEER (SElf-supERvised), a billion-parameter self-supervised computer vision model that can learn from any random group of images on the internet. // ai.facebook.com

# Coronavirus

Takeaways from one year of lockdown // mingyuan, 6 min

Covid 3/4: Declare Victory and Leave Home // Zvi, 16 min

Attempt at a gears model of getting infected with COVID-19 // ChristianKl, 1 min

What other peptide vaccines might it be useful to make? // jacobjacob, 1 min

How do you run a fit-test for a mask at home when you don't have fancy equipment? // ChristianKl, 1 min

# Instrumental

Seven Years of Spaced Repetition Software in the Classroom // tanagrabeast, 41 min

Texas Freeze Retrospective: meetup notes // jchan, 13 min

Enabling Children // lincolnquirk, 4 min

Check OK, babble-read, optimize (how I read textbooks) // AllAmericanBreakfast, 6 min

Tap Water and Filtration // manaterra, 1 min

# Epistemic

We should not raise awareness // lsusr, 3 min

The Flexibility of Abstract Concepts // lsusr, 5 min

Noticing and Overcoming Bias // neel-nanda-1, 8 min

# Ai

Fun with +12 OOMs of Compute // daniel-kokotajlo, 15 min Favorite

The case for aligning narrowly superhuman models // ajeya-cotra, 46 min

MIRI comments on Cotra's "Case for Aligning Narrowly Superhuman Models" // RobbBB, 32 min

Takeaways from the Intelligence Rising RPG // quinn-dougherty, 14 min

Multimodal Neurons in Artificial Neural Networks // Kaj_Sotala, 1 min

# Meta-ethics

What ethical thought experiments can be reversed? // MathieuRoy, 1 min

Subjectivism and moral authority // joekc, 17 min

# Anthropic

Grabby aliens and Zoo hypothesis // avturchin, 1 min

# Decision theory

A whirlwind tour of Ethereum finance // cata, 13 min

Kelly *is* (just) about logarithmic utility // abramdemski, 15 min

A non-logarithmic argument for Kelly // Bunthut, 2 min

A thought on gym membership models and from there the whole world… // Curious Couch Potato, 10 min

An Org Games Example // Robin Hanson, 2 min

# Math and cs

A Semitechnical Introductory Dialogue on Solomonoff Induction // Eliezer_Yudkowsky, 69 min

How To Think About Overparameterized Models // johnswentworth, 7 min

Naïve Set Theory - Part 1: Construction of Sets // sai-sasank-y, 3 min

# Books

Book review: "A Thousand Brains" by Jeff Hawkins // steve2152, 23 min

Notes on "Bioterror and Biowarfare" (2006) // MichaelA, 4 min

# Ea

Limits of Giving // jkaufman, 1 min

When I left Google // jkaufman, 2 min

# Community

"If You're Not a Holy Madman, You're Not Trying" // abramdemski, 3 min

Project: Debiasing Politics via Crowdsourcing // Leonid, 2 min

Grading myself on SSC's 2020 predictions // knite, 1 min

Open & Welcome Thread – March 2021 // Yoav Ravid, 1 min

AI Safety Beginners Meetup (Pacific Time) // Linda Linsefors, 1 min

# Culture war

02/28/2021 - Myanmar Diaries; Context // rockthecasbah, 6 min

The Puce Tribe // mike_hawke, 6 min

Idea selection // krbouchard, 2 min

# Misc

I'm still mystified by the Born rule // So8res, 12 min

Economic Class // lsusr, 4 min

Are the Born probabilities really that mysterious? // adele-lopez-1, 1 min

How to end stagnation? // jasoncrawford, 4 min

Thoughts On Computronium // Darklight, 3 min

# Rational fiction

AIDungeon 3.1 // yair-halberstadt, 2 min

# Videos of the week

The promise of quantum computers | Matt Langione // TED, 12 min