New post: going into our investigation of the HF attack (before Black Hat), I was very wrong about what basically happened. This incident was far more serious than I expected, and far more serious than previous documented misalignment incidents. https://t.co/QWQxD2Q179
fable is suddenly telling me its thinking about ethical concerns with my request as the first thing it does on ~every prompt. does it think i'm a bad guy or is this a general thing?
@ericwind Beautiful watch and great story, but if an unknown rube like me I bought something like this someday I don't think the "we believe" would work as well coming from me as it does coming from you!
@TheZvi Longtime reader, raised Southern Baptist (small venn diagram!)
Religious teen me would say to stop at the point their next marginal effort would have a negative impact on your lifelong openness to conversion. That happens pretty quick (maybe 0.1% of lifetime interactions +1?)
@chrislakin hard to share much more without sharing MUCH more!
Short version: if you don't feel great but you're working on an instrumental goal (like a promotion), easy to imagine you don't feel great because you haven't succeeded yet.
Check enough boxes, and that mirage dissipates.
@robinhanson did you look at the images they used for these things in the paper?
not looking would be bad; looking and still posting this misleading summary might be worse.
@tszzl SFF and Lightcone Commons are a great way to fit this need! Lab employees can delegate allocations to recommenders who are focused on alignment and control.
@paulg@neilkli@ryangrim so it's irrational for someone to move from silicon valley to (other) now because of talent network, but your threat model is that enough people will make this stupid choice to create a critical mass of stupid choicemakers somewhere else, that it would then be great to join?