Two AI programs invented by Facebook created their own language suddenly while an experiment was ongoing. The experiment was canceled by the programmer immediately.
Metro: Facebook robot is shut down after it ‘invented its own language’
According to the media, these two AI are communication bots. They were in training with talk to each other. They used English as a way of communication. However, they began to invent new words which had not been programmed for mutual communication.
I am not sure these bots equip deep learning technique. Use of some abbreviated terms is frequently seen in the communication among people in the close relationship. Thus, inventing some fragments of words is not surprising for an AI. The issue is what was the intention of the AIs, or should I say, "Did these bots think by themselves in introducing a new language?"
If so, it is a quite serious issue. Whether an AI can acquire a consciousness has been broadly discussed. We will able to ask the AI why did it invent an own language. If the answer is naturally reasonable, this AI may have a mind.
On the other hand, I hardly believe this AI can answer this question. AI with a will would not be developed at least for a decade.
No comments:
Post a Comment