✳flâneur — a map of the web's best reading

Supplement to “Why AI alignment could be hard”

cold-takes.com · 6,149 words · saved by 1 readers

> This page is authored by Ajeya Cotra [https://www.openphilanthropy.org/about/team/ajeya-cotra]. This post is a supplement to Why AI alignment could be hard with modern deep learning [https://www.cold-takes.com/p/67757b1f-ddc7-4691-b94b-10ae84cea84d/]. In this post, I: * Give a more in-depth explanation of how deep learning works

This page is authored by Ajeya Cotra . This post is a supplement to Why AI alignment could be hard with modern deep learning . In this post, I: Give a more in-depth explanation of how deep learning works with more technical detail, though I still aim for it to be fairly accessible to a general audience ( more ). Illustrate how we might train powerful deep learning models to do open-ended real-world tasks like the ones involved in PASTA ( more ). How deep learning works This section gives a high-level not-too-technical introduction to how deep learning works. It’s coming largely from a theory p

Explore this link on the map →

Supplement to “Why AI alignment could be hard”

related reading