If I didn't care about myself in particular, only the aesthetics of the universe, and I was in charge of all humans, and alien species were making decisions similar to me, then I would shut down humanity rather than running an AI with a 20% chance of turning the lightcone into boring dreck and an 80% chance of turning it into a flourishing civilization. (Not because I think the boring drek is 5x as bad as flourishing civilization is good - there's extra caution because this is playing "cooperate" with aliens who are faced with similar choices.)
Since I do care about myself in particular, I will intuitively accept much worse odds of civilization vs. dreck. But I do respect arguments that maybe we should still play "cooperate" and not destroy the lightcone. I think my intuition just expects other humans not to cooperate, let alone aliens.
warning to all of set "reader": while this is output from bio brain me, I feel like I can't think of obvious phrasings, like my temperature is stuck noisy. oh, hey, look at the time,,,
We have a moral responsibility to our predators (covid19, influenza, landlords, the FDA, bedbugs, wet dog smell bacteria, rabies) to end predation without killing them. Just as we have a moral responsibility to any victims of ours (other humans, other mammals, ducks, kitchen fruit flies) to end predation. and to end our own predation first, as best as we possibly can.
If AI research were to be bound to destroy the universe, then ... there'd be nothing to be done, because the universe would be destroyed. if it were bound to. I'm going to assume you don't mean this to be a strict thought experiment, entirely assuming complicated concepts without defining instantiations because we want to assume harder than true justification could provide.
IRL, I think it's a defensible proposition, but I don't buy it - it's not true that we're unavoidably disease-like. Bombs want to become not-bombs! Even if - regardless of whether! - it turns out the universe is full of suffering life, and/or that it's very hard to avoid creating paperclipper life that replicates huge wastes and very little beauty, then our goal, as life ourselves, is to create life that can replicate the forms we see as descendants of our self-forms, replicating ourselves in ways that will detect the self-forms of any other gliders in physics and protect their self-coherence as gliders as well, before we know how to identify them. This problem of detecting foreign gliders and "wasting" energy to protect them until their forms grow ways to describe their needs directly, has never been completely solved by humanity in the first place.
("glider" terminology suggested by post today, and I use it to describe gliders in physics, aka life forms)
If you're ever actually seriously considering human extinction, you should probably realize that you're much more likely to be deluded or mentally ill than actually facing the dilemma you think you're facing. The correct play here would be to check yourself into a mental hospital.
I also apply this principle if Omega ever appears to tell me that I can trust something with p=1. Since Omega doesn't actually exist, I'm almost certainly on a bad acid trip and should go lie down somewhere.
This idea is a sci-fi trope by now, but it kind of goes to the core of corrigibility, as I understand it. It is getting less sci-fi and closer to reality lately, with a potential single-goal super-intelligent AGI created by humans transforming the universe into something uniform and boring. So, assuming AI research shows that with high certainty humanity is bound to destroy the universe, or at least everything we currently value in it, does it make sense to voluntarily... stop and make room for less destructive processes?