AI fear
mardi 30 mai 2023 à 10:17Here is an example of someone who became terrified of "Artificial Intelligence" because he saw some output from ChatGPT.
It is clear he believes ChatGPT understands some sort of _meaning_ in its output. In fact, we know it does not understand anything. It is incapable of having knowledge and incapable of having intentions.
I am not trying to claim that real artificial intelligence will be safe. I don't know what it might do. Perhaps we should take some precautions -- such as limiting access to facilities for making DNA sequences to order. Those facilities might be used for mass murder even by plain old humans.
But we can't learn anything about hypothetical future artificial intelligences by looking at doubletalk-generating systems such as ChatGPT.