Most of the great experiences people have discussed so far have been human powered high-touch interactions. Most of these services promise long-term AI automation without a clear path or the team to get there. To survive, many of these new companies have shifted to high fees to cover the cost of human labor. The broader and more complex the potential requests from customers, the more automation (and profitability) looks like a dream.
…ems. The input and output to a user are both text, so you can skip the complexity of voice parsing. More importantly, the async nature of messaging allows the opportunity for handling each response in a way that is dependent on the level of machine understanding. In other words: by talking with Siri, you quickly can tell Siri is a machine. It would be foolish for Apple to try and act like there’s a chance Siri is even part human. What if you were texting with Sherri though, and some messages were sent by a human and some by an algorithm? If the combined system of software+agent ensures that a user sent message is always well understood and that the system sent messages are always sensible responses to the user, do you think of Sherri as human or a machine? If Sherri’s communication skills are at least as good as a human, but she has the tools and data access of a powerful machine then we think Sherri is a super-human.