Rejected for the following reason(s):
- Insufficient Quality for AI Content.
- I think this is a reasonable question, and there might be more stuff to talk about this, but the framing here doesn't really give much traction on a version of the conversation that would move the discussion forward.
Read full explanation
So, I am not some well known philosopher, mathematician, computer scientist, or anything as "important" as some people here might do for a living.... but if I get an answer here, I'll start to divulge some more information about myself and why I'd like to start joining conversations surrounding alignment.
Despite this not being my main "field"... I have been developing a framework anyway for myself mainly, and thought I'd share. It's one that's trying to help us understand adversarial cognition as an epistemic failure mode. Basically, the way humans default to enemy models of thinking when they face complex systems (no matter what era of time we talk about) even when it degrades our own understandings.
Questions:
1. Since I'm new here, was wondering if LessWrong has discussed this before?
2. Is there work examining whether "safety" framing itself introduces bias into alignment research?
3. If you're working on alignment, what does non-adversarial alignment even look like? Is that even a coherent concept, or am I just a naive person who doesn't understand AI risk? I'm open to changing my mind on anything, but I honestly just don't see us heading into full on dystopia, I think we can have a better world going forward. But open to hearing all the reasons why I'm wrong for sure. Many I know are doomers.
4. Is it possible to take AI risk seriously while also questioning whether adversarial framing is the best model?
Anyway, if you read this, thank you! I didn't mean to be so self depricating towards the top, but I'm not going to delete it... just going to be myself.
I have a longer writeup if anyone is interested, but wanted to first even see if this direction seems useful or if I'm missing existing work that covers this ground? I'm guessing at some point there's been conversations surrounding Nonviolent Communication, Princapled Negotation, Bohmian Dialogue, etc. but can't find it.
-Ataraxis ... Maximus ;)