The only people still “unreservedly” pushing AI are either stupid or greedy, says Peggy Noonan in The Wall Street Journal. The technology’s own creators can’t explain exactly how it “does what it does”. And many of those who understand it best are terrified. Geoffrey Hinton, the British computer scientist sometimes called the “godfather of artificial intelligence”, says it’s “not inconceivable” that AI will wipe out humanity. Last month more than 1,000 tech leaders, including Elon Musk, signed a “briskly direct” open letter urging a temporary moratorium on any further development of advanced AI systems. They’re absolutely right. AI will ultimately be as benign or as malignant as its creators. And currently those creators are the Big Tech overlords – who have repeatedly shown themselves to be “not at all preoccupied” with how their work might harm others.
Pausing development isn’t anything like enough, says veteran AI expert Eliezer Yudkowsky in Time. The reason this technology is so dangerous is because the algorithm doesn’t care about human life – so if it’s asked to solve climate change, say, or even just to rid the world of spam email, it might decide that the most efficient solution is to wipe out the human race. And there is no plan for how we can avoid that. The only safe thing to do is “shut it down”, with “no exceptions, including for governments or militaries”. We need to track all the hardware that bad actors could use to make it, and be ready to “destroy a rogue datacenter by airstrike” if necessary. Because if we let AI development continue unchallenged, “everyone will die”.