andrewedstrom
Joined 9 years ago
Joined 9 years ago
In it, they have GPT-4 generate solutions to coding problems, but instruct it to insert backdoors into the solutions some fraction of the time. Then, they explore different ways to use a weaker model (GPT-3.5) to detect these backdoors. Pretty interesting.
The site is called Modern Serial, and it lets you read books from Standard Ebooks in 10 minutes a day as Substack-style email newsletters.
They’re weird organisms composed of multiple distinct animals that cannot survive independently and that originate from the same fertilized egg. But they are unambiguously multicellular.
Open source models are already being used for all kinds of nefarious purposes. Any safety controls on a model are easily stripped off once its weights are public.
Usually I love open source software. Most of my career has been spent writing open source code. But this is powerful and dangerous technology. I don’t believe that nuclear weapons should be open source and available to all either.
Many people in AI safety are young. She has more professional experience than many leaders in the field.
Thanks for everything you do!
There is no tax penalty for moving bitcoin. You should definitely move most of this elsewhere.
https://twitter.com/andrewedstrom https://andrewedstrom.substack.com/ andrewedstrom.com