As long as the input data doesn't stray too far away from what the human masters have told it to expect.
I believe in the potential value of AI; but it’s not going to be an easy thing and we have to do it carefully and thoughtfully. As necessary as I see development of AI, I greatly fear rushing into it.
Many AI scientists, including Musk, Wozniak, Bengio, signed an open letter a couple of years ago calling for a six month pause in development. They felt we needed some time to figure this whole thing out and come up with some standards and ‘guard rails’. I can’t help but believe that they had seen things that caused them concern:
https://futureoflife.org/open-letter/pause-giant-ai-experiments/