I know that probably I am talking a little too much about Artificial Intelligence, the fact is that I am overwhelmed by news involving Artificial Intelligence. This one is about a chatbox that is better than us in social behaviour, and area that -by definition- should be our own turf!
An international research team lead by prof. Iyad Rahwan from MIT has developed a chatbox based on an Artificial Intelligence algorithm (S# – read S Sharp) to make social interactions between humans and machines possible. Testing the chatbox it turns out that it performs better, in social terms than two people conversing with each other.
Reading their paper explaining their results is quite interesting. The chatbox has been “programmed” to be straight in its interaction, never lie. It turns out that in the tests human to human conversation went well about 50% of the time (the goal of the conversation was to find a compromise in a gaming situation), au pair with the conversation between a human and the chatbox. But 50% of the time the human to human was worse than the human to chatbox conversation and the reason was that sometimes in trying to reach a compromise humans … lie!
This is also telling us the importance of a good “education” and morality for robots. My suspicion is that if they were taught to lie, they would be better than us also in lying…
It is not just this news. In the last week I had the opportunity of discussing with colleagues at EIT Digital on proposals of research themes for our Doctoral School coming from the industry in partnership with Milan Polytechnic and two of them were on the use of chatboxes to help interactions with children with mental disabilities and another to help communication between human workers and robot workers in the framework of Industry 4.0.
AI is dangerously entering into our private sphere, something that just in the last century (but that was less than 20 years ago) seemed impossible. And all signs are pointing towards an even faster evolution. Fasten your seat belt for a bumpy ride.