I’d invoke Ender’s Game here: If you’re unfriendly, no amount of communication will ever resolve that gulf, and either you or humanity HAS to be destroyed in the end.
Or we could go the biology route: I don’t test a potential new virus on human subjects until AFTER I know it’s safe.
Or we could just go with that I don’t care about the AIs story, because the question of “can I prove you’re friendly” is so vastly much more important.
I would assume there is quite a bit of medical science devoted to exactly that question. I know trials on similar animals were common for a time (mice seem to still be common), and we can also just do various chemical extrapolations. I’m not a medical student so I really wouldn’t know the nuances, but I’ve read enough to know there are various trials before you start on human experiments.
Drugs routinely fail in human trials, and one would guess that the drug companies did not start the human trials knowing that the drug killed all the mice or something. So we know that even if the animal trials go perfectly, it’s quite likely they’ll be unsafe in humans.
(What’s that quip again? “We can cure anything in mice except the cures not working on humans”? Sure I saw something like that somewhere.)
A while back I read a Slate article claiming that laboratory mouse strains have over time become optimized for experimental sensitivity rather than for suitability as model organisms, and that this leads to a number of medical fragilities that humans don’t have. Particularly in terms of cancer research: the kinds of tumors that pop up in highly inbred short-lived mice apparently don’t have a lot in common with human-typical cancers.
I haven’t read enough in this area myself to have a very informed opinion, and I can’t find much actual research beyond what Slate cites, but it sounds plausible.
Oh, I know the process is hardly perfect. My point is simply that we can gain a lot of information BEFORE we start human trials. And, as a society, we’ve concluded that it’s generally stupid to skip that precaution.
I’d invoke Ender’s Game here: If you’re unfriendly, no amount of communication will ever resolve that gulf, and either you or humanity HAS to be destroyed in the end.
Or we could go the biology route: I don’t test a potential new virus on human subjects until AFTER I know it’s safe.
Or we could just go with that I don’t care about the AIs story, because the question of “can I prove you’re friendly” is so vastly much more important.
(AI DESTROYED. You bugger lover ;))
How, exactly, do you determine if a virus is safe for humans without using humans?
I would assume there is quite a bit of medical science devoted to exactly that question. I know trials on similar animals were common for a time (mice seem to still be common), and we can also just do various chemical extrapolations. I’m not a medical student so I really wouldn’t know the nuances, but I’ve read enough to know there are various trials before you start on human experiments.
Drugs routinely fail in human trials, and one would guess that the drug companies did not start the human trials knowing that the drug killed all the mice or something. So we know that even if the animal trials go perfectly, it’s quite likely they’ll be unsafe in humans.
(What’s that quip again? “We can cure anything in mice except the cures not working on humans”? Sure I saw something like that somewhere.)
A while back I read a Slate article claiming that laboratory mouse strains have over time become optimized for experimental sensitivity rather than for suitability as model organisms, and that this leads to a number of medical fragilities that humans don’t have. Particularly in terms of cancer research: the kinds of tumors that pop up in highly inbred short-lived mice apparently don’t have a lot in common with human-typical cancers.
I haven’t read enough in this area myself to have a very informed opinion, and I can’t find much actual research beyond what Slate cites, but it sounds plausible.
Oh, I know the process is hardly perfect. My point is simply that we can gain a lot of information BEFORE we start human trials. And, as a society, we’ve concluded that it’s generally stupid to skip that precaution.