One particular way that “aligned” AI systems could make things work is if they accidentally “corrupt” our values
Did you mean “worse” instead of “work” here?
these should be taken as disagreements on how to solve these problems, not a disagreement that the problems exist.
I’m definitely not too attached to my proposed angles of attack either, and mainly wanted to give some ideas as existence proof that there are things that can be done from a technical perspective.
I’m not sure why we’re focusing on value corruption in particular. [...] I don’t have a great answer to the problem of competing aligned superintelligent AI systems.
No, I just didn’t get to that comment by the time the newsletter was sent out. (I’ve been a bit busy, and iirc that comment was either really close to or after the newsletter release time.)
No, I just didn’t get to that comment by the time the newsletter was sent out.
Ah, ok. Assuming you’ve changed your mind on that topic or had your questions satisfactorily answered, I wonder if there’s a way to update your newsletter readers without being too awkward or taking up too much time or space in the newsletter. (I’d guess that most of your readers aren’t following the discussions here.) Is that something you’ve thought about before?
Did you mean “worse” instead of “work” here?
I’m definitely not too attached to my proposed angles of attack either, and mainly wanted to give some ideas as existence proof that there are things that can be done from a technical perspective.
I thought I gave pretty reasonable answers at https://www.lesswrong.com/posts/HTgakSs6JpnogD6c2/two-neglected-problems-in-human-ai-safety#dykZxAXGr6sk7bh4e. Do you disagree with what I said?
Yes, fixed.
No, I just didn’t get to that comment by the time the newsletter was sent out. (I’ve been a bit busy, and iirc that comment was either really close to or after the newsletter release time.)
Ah, ok. Assuming you’ve changed your mind on that topic or had your questions satisfactorily answered, I wonder if there’s a way to update your newsletter readers without being too awkward or taking up too much time or space in the newsletter. (I’d guess that most of your readers aren’t following the discussions here.) Is that something you’ve thought about before?
Yeah, I do have a way—adding it to a subsequent newsletter after the highlights section. You’ll see that this week.