That being said, anyone who wants to stop all development is living in fantasy land. If we don't develop it, other countries will.
AI is easy to stop on a technical level.
Data Centers for training have to be huge, and have to pull huge amounts of power, and generate huge amounts of waste heat. They can be spread out for usage, but can't be distributed for training, since you run into speed of communication limitations. It requires huge amounts of specific computer components that sit at the peak of what humanity can develop, and it takes decades to build the facilities to start building the chips you need. There's only a handful of countries in contention to make AI, and bombing people into the stone age actually stops it.
AI far easier to stop than Nuclear Weapons. The US announcing (and following through) that it'll go to war and be willing to drop cruise missiles or nukes on anyone trying to develop AI would easily hold back AI development indefinitely. We just don't have the political will to do it, because nobody in power seriously thinks that the Singularity is going to kill everyone.
Unfortunately I don't think it's going to stay this way for very long. AI training will get more efficient, needing less data and hardware resources. Plus hardware will steadily get better. But we should be trying to nip AI in the bud while it is still relatively young, including at least trying to stop the necessary hardware being acquired.
We would likely deal with it the same way we did with nuclear proliferation.
For countries that don't have nukes and try to get them, we use economic pressure, then sabotage (Stuxnet), then assassinations, then actual bombs. And we have done all of those steps several times.
We almost bombed China to stop them from getting nukes, and arguably the only reason they have nukes today is because they had Jewish support and espionage feeding them support while stopping our response.
AI is easy to stop on a technical level.
Data Centers for training have to be huge, and have to pull huge amounts of power, and generate huge amounts of waste heat. They can be spread out for usage, but can't be distributed for training, since you run into speed of communication limitations. It requires huge amounts of specific computer components that sit at the peak of what humanity can develop, and it takes decades to build the facilities to start building the chips you need. There's only a handful of countries in contention to make AI, and bombing people into the stone age actually stops it.
AI far easier to stop than Nuclear Weapons. The US announcing (and following through) that it'll go to war and be willing to drop cruise missiles or nukes on anyone trying to develop AI would easily hold back AI development indefinitely. We just don't have the political will to do it, because nobody in power seriously thinks that the Singularity is going to kill everyone.
Unfortunately I don't think it's going to stay this way for very long. AI training will get more efficient, needing less data and hardware resources. Plus hardware will steadily get better. But we should be trying to nip AI in the bud while it is still relatively young, including at least trying to stop the necessary hardware being acquired.
We would likely deal with it the same way we did with nuclear proliferation.
For countries that don't have nukes and try to get them, we use economic pressure, then sabotage (Stuxnet), then assassinations, then actual bombs. And we have done all of those steps several times.
We almost bombed China to stop them from getting nukes, and arguably the only reason they have nukes today is because they had Jewish support and espionage feeding them support while stopping our response.