 |
| from here |
Computer scientists have done it again. Another artificially intelligent racist. Apparently not enough people learned the lesson of Tay. You can't train an AI on unchecked input from arbitrary people on the Internet and expect to wind up with something worthwhile. The assholes among them will do everything they can to spoil it.
I know input validation is probably not the right terminology for the AI context, but in the broader sense the chatbot is a program and malicious input is malicious input.
0 comments:
Post a Comment