24.9.22

The Trolling Test

 It seems appropriate that we adapt our expectations of conversational AI to the most salient problems of our current times. In this vein, I would say that the “trolling test” is the most appropriate test of a conversational AI bvo


bot’s. effectiveness. Instead of evaluating these programs in isolation, why not include a real=-world trial, making use of a public service such as Twitter?

No comments:

Post a Comment