Links for 2023-06-01
Improving Mathematical Reasoning with Process Supervision: https://openai.com/research/improving-mathematical-reasoning-with-process-supervision
We've trained a model to achieve a new state-of-the-art in mathematical problem solving by rewarding each correct step of reasoning (“process supervision”) instead of simply rewarding the correct final answer (“outcome supervision”). In addition to boosting performance relative to outcome supervision, process supervision also has an important alignment benefit: it directly trains the model to produce a chain-of-thought that is endorsed by humans.
The death of self-driving cars is greatly exaggerated https://www.understandingai.org/p/the-death-of-self-driving-cars-is
🧠👁️ MindEye: A new state-of-the-art fMRI-to-image approach that retrieves and reconstructs images from brain activity. https://medarc-ai.github.io/mindeye/
PaLM-2 & GPT-4 in "Extrapolating GPT-N performance" https://www.lesswrong.com/posts/75o8oja43LXGAqbAR/palm-2-and-gpt-4-in-extrapolating-gpt-n-performance
Using Photoshop Generative Fill Beta to expand Midjourney photos: https://www.reddit.com/r/midjourney/comments/13w4miy/using_photoshop_generative_fill_beta_to_expand/
Probabilistic AI that knows how well it’s working https://news.mit.edu/2023/probabilistic-ai-knows-how-well-its-working-0525
Purely Passive Learning Can Allow An Agent To Learn Generalizable Strategies For Using Causal Structures https://arxiv.org/abs/2305.16183
AdaPlanner: Adaptive Planning from Feedback with Language Models https://arxiv.org/abs/2305.16653
58% of US adults have heard of ChatGPT; 14% have tried it. https://www.pewresearch.org/short-reads/2023/05/24/a-majority-of-americans-have-heard-of-chatgpt-but-few-have-tried-it-themselves/
“Six years of intensive formalisation activity seem to show that even the most advanced results, drawing on multiple fields of mathematics, can be formalised using the tools available today.” https://arxiv.org/abs/2305.14407
Unreal Engine 5: Next Level Games Are Coming! https://www.youtube.com/watch?v=dVgx3uJuHOE
In defense of hereditarianism: a reply to Friedman https://www.emilkirkegaard.com/p/in-defense-of-hereditarianism-a-reply
James Lee systematically dismantles the claims made by IQ "environmentalist" Richard Nisbett in his book Intelligence and How to Get It. http://laplab.ucsd.edu/articles2/Lee2010.pdf
China, India Kick Out Nearly All of Each Other’s Journalists as Rivalry Escalates [Wall Street Journal] https://archive.is/xGYq8
Fewer than half of Gen Z men think “feminism has made America a better place”: https://twitter.com/BradWilcoxIFS/status/1663253365902409752