Should Chat-bots be human-like?
To identify machine intelligence, Turing developed in 1950 the so-called Turing Test. The idea is quite simple, different experts are communicating text-only with two players and have to distinguish, who of them is a human being and who a machine. Up to today, machines had not been able to convince all the human judges, but results get better. In 2014 a chat-bot software portrayed a 13-year Ukrainian boy and with this could convince in the Turing Test 30% of judges, that it was a real boy.[ IT companies assume that after the development of the smart-phone, intelligent chat-bots will be the next important step. Such software can connect to global databases, connect and interpret them.
But should intelligent chat-bots be human-like? Humans prefer a bot which pretends to be human or they would perceive the same benefit from a bot, which clearly acts like a machine?
There is an difference between the underlying AI and the interface. A Chat-bot is an interface. The choice of interface should be based on an analysis of the audience and their needs.
36 months ago
Ed Addison made a valuable point. We need to differentiate between the interface and the technology behind it. A chat-bot technology framework can be wrapped into Alexa or Sophie or skype contact. What makes a machine look Human like is largely dependent upon how we human perceive. Another aspect is evolution. I have seen transition from no mobile to smartphone, very difficult and conflicting change. The generation born in smartphone age would not have that issue, so we adopt to the technology.