The program for learning effective conversation

Dr. Alex Pentland and his colleagues at the Massachusetts Institute of Technology have developed a very interesting program that will be useful for improving meetings or group podcasts, when interlocutors often interrupt each other, speak with different volume or someone from the speaker suppresses others with your intellect. The new development can analyze the conversation with the participation of different people and report to each participant about the above-listed shortcomings in his way of leading the discussion.

True, not everything is so simple. For normal, a multi-week or even months-long adjustment to a specific person is needed. The system consists of a badge with sensors and a smartphone for recording information. A person should wear such a kit on himself for several weeks, and the program records absolutely all of his words, analyzing the tone, frequency and energy of the voice, as well as the speed of speech. Thus, a profile per person is compiled (Dr. Pentland calls this process “reality mining”, by analogy with “data mining”).

Over time, information accumulates, and the program is already analyzing specific situations and notices if the “host” begins to behave atypically or negatively affects the interlocutor. If a person has serious problems with communication, he can contact a specialist who will receive comprehensive information about the problem from the same system.

Dr. Pentland has been experimenting with similar technologies for a long time. For “reality mining”, he also uses infrared sensors (so that the system can identify when a person is opposite to another) and accelerometers for recording gestures. Some of the results of his work, the doctor published in the book “Honest Signals”, which was recently published by MIT Press.