00:00Roman, what do you think about J.D. Vance's comments? Should it be regulated? Can it be regulated safely? Is it just too broad?
00:11Right. So we're setting up this binary system where either you're going to make profits and grow economy or you have to be careful with technology.
00:20As I said before, existing narrow systems for solving specific problems will allow us to capture most of the financial benefit.
00:28You don't have to have a general superintelligence to solve specific problems.
00:33We have excellent examples. For example, protein folding problem was solved with a specific tool.
00:38It was not a tool developed for do everything for me. We can do the same for self-driving cars.
00:44We can do the same for specific diseases and so on. So that's a very important difference.
00:49Nobody's suggesting that technology is bad or AI is not great. I'm an AI researcher. I use AI every day.
00:56I just don't want to switch from comfortable tools which make me more productive to something nobody can control.
01:04And the problem with applying regulation to only specific deployments is that how do you enforce it?
01:12If you create a model smarter than me and capable of doing everything,
01:17how do you make it that no one uses it outside of specific domain of spell checking?
