“Can Robots Read Our Minds” Dr Alessandro Vinciarelli - TEDxUniversityofGlasgowSalon

Okay, I hope anybody can hear me. We are trying to answer in the few minutes we have at this position about this question; can a robot - or in general, a machine - read our mind? We start with a little game. You see a picture here; can anybody tell me what these two people are doing? Discussing, talking - what are they doing?

[AUDIENCE]

Shouting, arguing; any other?

[AUDIENCE]

Singing?!

[AUDIENCE]

If I ask you the type of relationship they have - friends, colleagues, student and supervisor - what type of-

[AUDIENCE]

Lovers singing!

The funny thing of this game is that there is no right or wrong answer here. What is interesting is that the picture is very poor: We don't know these people are. We cannot hear their voices. We don't know what they say. We cannot see their faces. And still we managed to give, confidently, a very high-level interpretation of the picture, in terms of social and psychological phenomena. So how this can happen? In general, our attention tends to focus pretty much on what people say, but there is a lot of nonverbal communication going on whenever there is social interaction. 

Here you see examples of what we call nonverbal behavioural cues. We as humans are very much sensitive to this cues. We perceive them and we interpret them in terms of social signals; so in terms of the relational messages that people exchange with one another. Let me go a bit deeper into this phenomenon, because it is really at the core of the answer we are going to give to our question.

When we talk about nonverbal behavioural cues we talked about essentially three elements; the first are the nonverbal behavioural cues that we have just seen in the picture. Here you have a list; it is very long, but it is not exhaustive. There are so many nonverbal behavioural cues that the psychologists have grouped them into five major classes - The Codes - and this is the second important element of nonverbal communication. Let's go one by one:

I think we all heard at a certain point in life that "It is not important how you are outside; it is important how you are inside." It's false. It has been investigated extensively; if you look nice, if you look good, your life is going to be much easier.

Here we come to the third important element of nonverbal communication. Nonverbal communication is a natural phenomenon; and like everything in nature, it has a function. Here we have a list of the functions that the psychologists have identified; some of the most important. Let's go by them one by one:

This is a very general introduction to nonverbal communication; now the point is, what has this to do with the question at the centre of this talk; what have this to do with the question whether a robot or a machine can read our mind? Well, it has to do in this way; what does it happen to this if I replace the person on the right with a machine? A machine means anything - any apparatus that has "sensors" - and a minimum of processing power. So the technological questions are, can I automatically detect nonverbal behavioural cues and can I automatically make sense of nonverbal behavioural cues?

The answer is yes in both cases; technology produces nowadays technological approaches that can detect the nonverbal and behavioural cues that have been mentioned, and they can make sense automatically of them in terms of social & psychological phenomena. Exactly the same way as we have been doing - when we have seen the picture at the beginning of this talk and we've figured out - we have guessed that those two people are singing, laughing, arguing, were being lovers and so on.

So when we go to the question we posed at the beginning - can a robot read our mind? - the answer is yes. Why it is so? Very simply because nonverbal communication is the physical, machine-detectable evidence of social and psychological phenomena. When we think of our inner life - of our mind, of our psychology - we tend to think of it as something abstract; something that we cannot really access with our senses. This is true; however the same inner life - the same psychology, the same mind - leaves traces in the physical world in terms of nonverbal behavioural cues, and here is the key for the machines to access it. If we can perceive the cues with our senses, then we can detect them with sensors and then we can understand that with artificial intelligence approaches.

Now, what is very much cool for the people like me that work on this type of technologies is that it is about people as much as it is about machines. It is very much interesting, you don't need only to know computer science, machines, mathematics etc; you also need to learn a lot about people, about psychology, about human sciences. What is a bit frightening about this technologies is that nowadays the progress is mostly machine-oriented; what does it mean? In this moment, we exploit the knowledge of we have about people to build machines that are more and more powerful at doing what we do every day - understanding and reading the mind of others - irrespectively of any effect that this can have on the life of people. In a sense, we are exploiting only one side of the lecture; we are learning that we cannot build a social intelligence machine if we do not understand people. So now the progress is machine oriented; we exploit the knowledge we have of people simply to build better machines. It would be nice to do it the other way around.

We cannot understand people if we cannot build social intelligence machines. What does it mean? These machines are a tool of unprecedented power to understand how people work; and it is very much an important to understand how people work, because if you understand it you can improve their life substantially. Let me give you a couple of examples that involve the school of computing science where I work. We are using a robot to teach autistic adults how to develop better social skills. This is crucial for them to make their life happier and healthy; in this way we better understand autism works, and through that we manage to improve the life of these people. Similarly, we use these technologies to detect as early as possible when children are mistreated by their parents. Why is this important? If you understand it early enough, you can increase a lot the chances of saving these children and making their life as adults happier and healthy.

So from this point of view we are on a very nice way, because we have the possibility to make technology that improve the life of people; but for what concerned § both the positive sides of these technologies, and the more worrying side of this technology? We should never forget that when we talk about artificial intelligence - because this is what we are talking about -"artificial" should be that big and "intelligence" should be that small. There is still a lot of work for us that work on improving the life of people for this; and still we don't need to worry much about the very negative effects, that potentially these technologies can have. Thank you very much.

--- END OF TRANSCRIPT ---

 
 

-VIEW MODE-


“Can Robots Read Our Minds” Dr Alessandro Vinciarelli - TEDxUniversityofGlasgowSalon

00:00:00