There’s an accidental harm article, which is meant to cover the risk of causing harm as an unintended effect of trying to do good, as discussed e.g. here. What you describe is somewhat different, since the risk results not so much from “attempts to do good” but from the development of a technology in response to consumer demand (or other factors driving innovation not directly related to altruism). Furthermore, misuse risk can involve deliberate attempts to cause harm, in addition to unintended harm. I guess all of these risks are instances of the broader category of “downside risk”, so maybe we can have an article on that?
I think there are indeed overlaps between all these things.
But I do think that the application of these terms to technological risk specifically or AI risk specifically is important enough to warrant its own entry or set of entries.
Maybe if you feel their distinctive scope is at risk of being unclear, that pushes in favour of sticking with the original AI-focused framing of the concepts, and maybe just mentioning in one place in the entry/entries that the same terms could also be applied to technological risk more broadly? Or maybe it pushes in favour of having a single entry focused on this set of concepts as a whole and the distinctions between them (maybe called Accident, misuse, and structural risks)?
I also wouldn’t really want to say misuse risk is an instance of downside risk. One reason is that it may not be downside risk from the misuser’s perspective, and another is that downside risk is often/usually used to mean a risk of a downside from something that is or is expected to be good overall. More on this from an older post of mine:
One definition of a downside is “The negative aspect of something otherwise regarded as good or desirable” (Lexico). By extension, I would say that, roughly speaking, a downside risk is a risk (or possibility) that there may be a negative effect of something that is good overall, or that was expected or intended to be good overall.
Also, I think I see “accidental harm” as sufficiently covering standard uses of the term “downside risk” that there’s not a need for a separate entry. (Though maybe a redirect would be good?)
There’s an accidental harm article, which is meant to cover the risk of causing harm as an unintended effect of trying to do good, as discussed e.g. here. What you describe is somewhat different, since the risk results not so much from “attempts to do good” but from the development of a technology in response to consumer demand (or other factors driving innovation not directly related to altruism). Furthermore, misuse risk can involve deliberate attempts to cause harm, in addition to unintended harm. I guess all of these risks are instances of the broader category of “downside risk”, so maybe we can have an article on that?
I think there are indeed overlaps between all these things.
But I do think that the application of these terms to technological risk specifically or AI risk specifically is important enough to warrant its own entry or set of entries.
Maybe if you feel their distinctive scope is at risk of being unclear, that pushes in favour of sticking with the original AI-focused framing of the concepts, and maybe just mentioning in one place in the entry/entries that the same terms could also be applied to technological risk more broadly? Or maybe it pushes in favour of having a single entry focused on this set of concepts as a whole and the distinctions between them (maybe called Accident, misuse, and structural risks)?
I also wouldn’t really want to say misuse risk is an instance of downside risk. One reason is that it may not be downside risk from the misuser’s perspective, and another is that downside risk is often/usually used to mean a risk of a downside from something that is or is expected to be good overall. More on this from an older post of mine:
Also, I think I see “accidental harm” as sufficiently covering standard uses of the term “downside risk” that there’s not a need for a separate entry. (Though maybe a redirect would be good?)