Q&A: Artificial Intelligence
Artificial Intelligence
Question
Hard to believe, but true: computers trained by Facebook to conduct negotiations using machine-learning artificial intelligence programs actually showed excellence at the task.
The surprise: it turned out they had begun developing a language of their own.
Answer
This should be discussed on two levels:
1. The principled-philosophical level: does this threaten the definition of the human being as a unique creature? Here, it seems to me that not much is new. It is impressive technology, but no more than that. Computers have long known how to do things that human beings cannot do. So now they know how to do one more thing that human beings can do as well: converse.
2. An existential threat to humanity (the computers will take over and become smarter than we are). Here too, I am not especially alarmed. A nuclear missile aimed at us is a more serious threat, and there is less we can do against it. So at most this is still just a technology that might be able to bother us—but how is that different from any other technological or other threat?