I was looking at this website while working on another project today, and checking their blog led me to this article, which while promotional for a company called inbenta, asked what I thought was a pretty good question. Now the answer they give - 32% - feels a bit like it was plucked from the air. But the discussion surrounding it about the metrics involved in even thinking about what would constitute a 'self-service rate' was worthwhile. There's another, more general post from last year on key metrics for evaluating chatbot performance. Here's another article from a different company (REVE Chat) on the same topic. And another (LivingActor). These discussions are note theoretical; chatboats are being tested in education now, as Sofy Carayannopoulos's description of the BU111 trial at Wilfrid Laurier University makes clear (pp. 112-127 of this 489 page PDF).
Today: 5 Total: 101 [Direct link] [Share]
