4 ms·
Or, we try to set common rules before the thing goes out of hand. We already have UN rules about fire weapons, and your argument applies to those as well. But j
by GTP 3y ago
Or, we try to set common rules before the thing goes out of hand. We already have UN rules about fire weapons, and your argument applies to those as well. But just as we have widely agreed upon rules for weapons, if there's the chance of AI being used used for war (which indeed seems to be the case, see e.g. UAVs) then we should regulate it as well. This leaves out civil uses of AI, but as we have safety regulations for every other technology, I don't see why AI should be an exception. Then we can discuss which regulations make sense and which not, but this is a different topic.
PS: I don't thik AI will pose an existential risk to our species anytime soon, if ever. But it can already cause damages if used improperly, like any other technology. So it has to be regulated, like any other technology.
- FrustratedMonky 3y agoOn one hand, I agree. Nuclear weapons are a risk to everyone, so the world(US, Russia) did come together in various treaties to limit production, development, etc... So, this could apply to AI also, everyone sees the risk and comes together to curb their us and development. On other hand: Nukes are very difficult to make. Iran and North Korea do not abide by the UN resolutions, they are continuing trying to develop them. We are just lucky that is not so easy. With AI, the barrier to entry is much lower. You can find things that can be weaponized for free on github. It isn't like they are going to bother reading the licenses and not use them. And for compute power, sure, they have harder time getting the best latest CPU's, and there is an embargo. But still, I think, far easier to get illegal gotten CPU's, than developing Nukes, the barrier is still lower.