Home Artists Posts Import Register

Content

TL;DR: I'm doing a Q+A, post questions!

In the quest for ways to make good content regularly and quickly, people often suggest doing livestreams to answer questions, and that always made me nervous because I wouldn't be able to do any editing or fact checking. But why not do a recorded Q+A?

So, post your questions here (or by Patreon message)! Mostly I'm looking for AI Safety questions obviously, but you can also ask about more general AI topics or about the channel. Probably not all questions will be a good fit for a video, but anything I don't answer in the video I'll answer here or in private message. Try to keep questions short so I can read them out easily, and let me know if you don't want your name included. I'll prioritise questions from Patrons, but I'll probably ask on Twitter too (and depending on the response maybe look through past YouTube comments as well), to make sure I have enough to fill a video.

I think that's everything... Ask away!

Comments

Anonymous

Hi Robert, Some AI researchers signed an open letter to ban Lethal Autonomous Weapons. What is your position on this issue? I personally do believe that soon any dependence on human decisions in the operation of all weapon systems would render them useless. The day is near when wars will be won in an hour, and with no human casualties. And this is good, I believe. But... while I am not afraid of SkyNet, it is quite scary to think that one day robots will do everything for us, and powerful humans will not need to control other humans in order to get what they want.

Jeremy

Would you put your brain in a robot body, and if so which side would you then be on in a Human vs. Robot war?

Jeremy

a (probably) more serious question: Could redirecting the attention of an AGI be used to mitigate damage, and conversely how do you avoid spending time on trolls / unimportant tasks? i.e. how does AGI "focus" on what it wants to achieve while still keeping it's essential resources available - and can that be reliably interrupted?