At this point, I have no idea what to believe. I don’t know if this is the case of the doomiest voices being the loudest, while the world is actually populated with academics, programmers and researchers who form the silent, unconcerned majority—or whether we genuinely are all screwed.
My sense is that this is broadly true, at least in the sense of ‘unconcerned’ meaning ‘have a credence in AI doom of max 10% and often lower’. All the programmers and academics working on AI presumably don’t think they’re accelerating or enabling the apocalypse, otherwise they could stop—these are highly skilled software engineers who would have no trouble finding a new job.
Also, every coder in the world knows that programs often don’t do what you think they’re going to do, and that as much as possible you have to break them into components and check rigorously in lab conditions how they behave before putting them in the wild. Of course there are bugs that get through in every program, and of course there are ways that this whole picture could be wrong. Nonetheless, it gives me a much more optimistic sense of ‘number of people (implicitly) working on AI safety’ than many people in the EA movement seem to have.
My sense is that this is broadly true, at least in the sense of ‘unconcerned’ meaning ‘have a credence in AI doom of max 10% and often lower’. All the programmers and academics working on AI presumably don’t think they’re accelerating or enabling the apocalypse, otherwise they could stop—these are highly skilled software engineers who would have no trouble finding a new job.
Also, every coder in the world knows that programs often don’t do what you think they’re going to do, and that as much as possible you have to break them into components and check rigorously in lab conditions how they behave before putting them in the wild. Of course there are bugs that get through in every program, and of course there are ways that this whole picture could be wrong. Nonetheless, it gives me a much more optimistic sense of ‘number of people (implicitly) working on AI safety’ than many people in the EA movement seem to have.