Interesting. Is the phenomenological work to try to figure out what kind of agents are conscious and therefore worthy of concern or do you expect insights into how AI could work?
Both, although I mostly consider the former question settled (via a form of panpsychism that I point at in this post) and the latter less about the technical details of how AI could work and more about the philosophical predictions of what will likely be true of AI (mostly because it would be true of all complex, conscious things).
Also the “phenomenological” in the name sounded better to me than, say, “philosophical” or “continental” or something else, so don’t get too hung up on it: it’s mostly a marker to say something like “doing AI philosophy from a place that much resembles the philosophy of the folks who founded modern phenomenology”, i.e. my philosophical lineage is more Kierkegaard, Hegel, Schopenhauer, Husserl, and Sartre than Hume, Whitehead, Russel, and Wittgenstein.
“Formally Stating the AI Alignment Problem” is probably the nicest introduction, but if you want a preprint of a more formal approach to how I think this matters (with a couple specific cases), you might like this preprint (though note I am working on getting this through to publication, have it halfway through review with a journal, and although I’ve been time constrained to make the reviewers’ suggested changes, I suspect the final version of this paper will be more like what you are looking for).
I’ve been a bit busy with other things lately, but this is exactly the kind of thing I’m trying to do.
Interesting. Is the phenomenological work to try to figure out what kind of agents are conscious and therefore worthy of concern or do you expect insights into how AI could work?
Both, although I mostly consider the former question settled (via a form of panpsychism that I point at in this post) and the latter less about the technical details of how AI could work and more about the philosophical predictions of what will likely be true of AI (mostly because it would be true of all complex, conscious things).
Also the “phenomenological” in the name sounded better to me than, say, “philosophical” or “continental” or something else, so don’t get too hung up on it: it’s mostly a marker to say something like “doing AI philosophy from a place that much resembles the philosophy of the folks who founded modern phenomenology”, i.e. my philosophical lineage is more Kierkegaard, Hegel, Schopenhauer, Husserl, and Sartre than Hume, Whitehead, Russel, and Wittgenstein.
I think the former is very important, but I’m quite skeptical of the later. What would be the best post of yours for a skeptic to read?
“Formally Stating the AI Alignment Problem” is probably the nicest introduction, but if you want a preprint of a more formal approach to how I think this matters (with a couple specific cases), you might like this preprint (though note I am working on getting this through to publication, have it halfway through review with a journal, and although I’ve been time constrained to make the reviewers’ suggested changes, I suspect the final version of this paper will be more like what you are looking for).