Links for 2022-10-28
gwern: “We know synthetic non-natural data like fractal pretraining works surprisingly well (it's a simplicity/Turing-machine prior); is agency also pretrainable from 'synthetic non-agent-generated data' like generative modeling of raw physics? Possibly:” https://pastebin.com/U7quKPhu
“A Vision of Metascience: An Engine of Improvement for the Social Processes of Science” By Michael Nielsen and Kanjun Qiu https://scienceplusplus.org/metascience/index.html
“Common Sense Machines has built a neural simulation engine we call CommonSim-1...These interfaces enable us to create digital replicas of real-world environments or imagine new worlds as simply as taking photos or describing a scene in text.” https://csm.ai/commonsim-1-generating-3d-worlds/
After a lifetime of blindness, newly sighted can immediately identify human locomotion https://news.mit.edu/2022/newly-sighted-can-immediately-identify-human-locomotion-1018
Robots in Surgery: “Although the attainable gains are higher for lower-skilled surgeons, they use the robot the least.” https://github.com/elenatafti/JMP/blob/main/JMP_Ashtari_Tafti.pdf
How to Take Over the Universe (in Three Easy Steps) https://www.lesswrong.com/posts/DvQ7cYxhnrZtWngvW/how-to-take-over-the-universe-in-three-easy-steps
Bumblebees love to play with balls. https://sciencedirect.com/science/article/pii/S0003347222002366
Meta-analysis finds that the Flynn effect isn't on g — that is, it has more to do with specific abilities than general intelligence — and that the Flynn effect and the black-white IQ gap (which has more to do with g) have different causes. https://www.sciencedirect.com/science/article/abs/pii/S0160289613000226
Our Homo Sapiens Ancestors Shared the World With Neanderthals, Denisovans, and Other Types of Humans Whose DNA Lives on in Our Genes https://singularityhub.com/2022/10/21/our-homo-sapiens-ancestors-shared-the-world-with-neanderthals-denisovans-and-other-types-of-humans-whose-dna-lives-on-in-our-genes/
This is what a lot of people are missing. They are oftentimes only exposed to one recent result and conclude that AI is still a long way off from achieving X. But they don't realize that machine learning is progressing at a breakneck speed.
“Do not look at where we are. Look at where we will be two more papers down the line.” — Károly Zsolnai-Fehér