"The capabilities of AI models being this strong, combined with the fact that we don't know how to make them safe, should be a concern to all," Jason Hausenloy, a policy lead at the non-profit Center for AI Safety (CAIS), told ABC News.
end partial quote from:
After thinking about this AI models are sort of like a jet passenger plane. They might fly and be fine or one day they might crash and kill everyone on board. So, if you aren't thinking of them like a big passenger plane or a cruise ship that might crash and kill everyone on board then you are being very very naive.
What if an AI model went rogue and tried to kill everyone in the U.S. or another country by screwing up public water systems to make them unsanitary or downright fatal if people drank that water without boiling it? This might happen only late one night while everyone was asleep. But, would anyone detect the problem in time so people didn't die? Not sure.
No comments:
Post a Comment