Best of Twitter - Weeks of June 27 & July 4 2022

Aarya @gd3kr

built a quick app that uses gpt-3 to convert from English to RegEx so you don't have to waste time on stackoverflow: autoregex.xyz

Dwarkesh Patel @dwarkesh_sp

.@SBF_FTX explains what makes someone a good founder. Full episode: 📺: youtu.be/5Af0d2DbK5w 🎙: open.spotify.com/episode/2SXFTj…

chris beiser @ctbeiser

pretty wild how there’s social pressure on high schoolers/college students to set up half-real volunteer/nonprofit groups, and hype them up with no consideration for how long they last or accomplishments where, when adults do this, we call it “being a con man”

Moritz Helmstaedter 🇺🇦 @mh_lab

What makes us human, at the neuronal network level? We discovered a dramatic (10-fold) expansion of interneuron-to-interneuron networks in the human brain compared to mouse: just published in @ScienceMagazine @MpiBrain @maxplanckpress science.org/doi/10.1126/sc…

Jon M Jachimowicz @jonj

🚨 New WP 🚨 In late 2020 we gave 5,000 low-income US participants $2000, $500, or nothing ($2.5m in transfers), measured survey outcomes, & (for 43%) have access to bank info. What did we find? 1/ (w/ @AniaJaroszewicz, @julianjamison, & Oliver Hauser) 🔓papers.ssrn.com/sol3/papers.cf…

alewkowycz @alewkowycz

Very excited to present Minerva🦉: a language model capable of solving mathematical questions using step-by-step natural language reasoning. Combining scale, data and others dramatically improves performance on the STEM benchmarks MATH and MMLU-STEM. goo.gle/3yGpTN7

Jacob Steinhardt @JacobSteinhardt

In 2021, I created a forecasting prize to predict ML performance on benchmarks in June 2022 (and 2023, 2024, and 2025). June has ended, so we can see how the forecasters did:

bounded-regret.ghost.ioAI Forecasting: One Year InLast August, my research group created a forecasting contest to predict AI progress on four benchmarks. Forecasts were asked to predict state-of-the-art performance (SOTA) on each benchmark for June 30th 2022, 2023, 2024, and 2025. It’s now past June 30th, so we can evaluate the performance of the f…

Jacob Steinhardt @JacobSteinhardt

To give an idea of just how much SOTA exceeded forecasters' expectations, here are the prediction intervals for the MATH and Massive Multitask benchmarks. Both outcomes exceeded the 95th percentile prediction.

Far Left Kyle @FLKDayton

This was made using the DALLE-2 neural network to extend Michaelangelo's creation of Adam.

Jared (is in London) Tumiel @jnearestn

some mutations are more/less likely than random, independent of neg. selection (violating key piece of evolutionary logic that mutations are uniform random) and *maybe* this is controlled by epigenetic effect on DNA replication fidelity??? @drmichaellevin any thoughts?

Jared (is in London) Tumiel @jnearestn

prompt engineering works on people, and on yourself you can say to yourself "you are Tyler Cowen, known for your deep, insightful, and unconventionally interesting interview style, about to sit down and talk to X" and you will absolutely have the best conversation of your day

^ this is legitimately extremely useful. Do not ignore.

Compound248 @compound248

Zuck is back.

Nicole Barbaro @NicoleBarbaro

🧠 The tiniest known brain belongs to this tiny wasp, at just 7,400 neurons. Humans have 86 BILLION neurons for reference.