Robotics and AI are expect­ed to deal with many of the glob­al chal­lenges that we are fac­ing. Poverty, inter­na­tion­al con­flict, health prob­lems. And this means that they will leave the fac­to­ry floors and that they will come into the envi­ron­ments where we are most, in our homes, in our streets, in our schools, and in our hospitals.

Here’s an exam­ple of a typ­i­cal envi­ron­ment. And these envi­ron­ments are social. We are social ani­mals by nature, and we inter­act social­ly, and we obey social rules all the time. It tells us what is ​“nor­mal” and it tells us how to behave. So if we take this tech­nol­o­gy and put it in these envi­ron­ments, this tech­nol­o­gy will stand out if it does­n’t com­ply, if it does­n’t have social intelligence.

And when we are not social­ly intel­li­gent, we are seen as rude, or threat­en­ing, and peo­ple real­ly don’t like us. And that is the same when you use tech­nol­o­gy and put it in environments. 

Here’s an exam­ple of how we made com­put­ers more social­ly intel­li­gent, because how can we make com­put­ers empathize. It’s a feed­ing robot that helps par­a­lyzed peo­ple to eat. And we made actu­al­ly quite a sim­ple mod­ule that lis­tens to the con­ver­sa­tion at the table and then offers the food at nat­ur­al inter­vals dur­ing the din­ner, which allows a per­son to com­plete­ly par­tic­i­pate in the rit­u­al that din­ing togeth­er is.

So, we can do for sim­ple things. But what about more com­plex social sit­u­a­tions? You can see speed-dating. We made a lot of peo­ple speed-date, and then know­ing the out­comes we could make the com­put­er think about what were the para­me­ters that pre­dict­ed a match. And in this case it was the vari­a­tion in the dis­tance between the two daters. I impart this wis­dom to you. But it shows that we can deal with quite com­plex social situations.

And the core idea behind mak­ing machines social­ly intel­li­gent is that sen­tient beings need social ref­er­enc­ing to learn. We do this much more, for instance, than chim­panzees. We hold our babies, we look at them, we put them in a chair, and we talk to them, we wave at them, we show them objects. And babies emu­late our behav­ior and they learn. And it’s thought that this social ref­er­enc­ing leads to empa­thy and know­ing the­o­ry of mind, and even enculturation.

Enculturation allows us much more effec­tive­ly, for instance than chim­panzees, to teach chil­dren how to use that the think­ing tools that we all have in our brain very effec­tive­ly. So just to com­pare it takes a chim­panzee child about sev­en years to learn to hit a nut with a rock.

So, social ref­er­enc­ing is so great robots should do it, too. But first there are some tech­ni­cal chal­lenges that we need to solve. For instance, low ener­gy con­sump­tion. Throughout these tech­ni­cal chal­lenges, what these robots real­ly need to do is under­stand the social envi­ron­ment that they are in. 

So, this robot, a SPENCER robot at Schipol air­port, it needs go around a lit­tle fam­i­ly and not dri­ve through them. And here you see the FROG robot. It’s an out­door guide robot, and this robot has to respond nat­u­ral­ly peo­ple. It has to behave nat­u­ral­ly. And you see a blind per­son try­ing to under­stand FROG and FROG try­ing to under­stand the blind per­son. The mod­el that FROG has of it users must also deal with impairments. 

And what hap­pens is when we put these behav­iors into robots, they are eas­i­er to under­stand because they are much more famil­iar. But it also means that they are becom­ing very rich social char­ac­ters. And it may even mean that we empathize with them or maybe even devel­op a social bond with such robots.

This is the EASEL robot, and you can see a boy real­ly engag­ing with this robot even though one of the huge chal­lenges, real­ly these tech­ni­cal chal­lenges that I told you about to get these robots run­ning for long peri­ods of time, a very large chal­lenge, is our own fear. Because we are not quite sure what hap­pens to peo­ple when they start relat­ing to robots, and if they change how they relate to people.

So it’s extreme­ly impor­tant that we design and devel­op these robots to fit the role that they have to do. Here you see a TERESA telep­res­ence robot that elder­ly peo­ple can use to par­tic­i­pate at a dis­tance. So they log into the room from a remote loca­tion. And over time the arti­fi­cial intel­li­gence takes over their posi­tion­ing behav­ior so they can focus on socializing.

These robots need to seam­less­ly inte­grate into our envi­ron­ments. But they are huge­ly inte­grat­ed machines them­selves. They need machine learn­ing, com­put­er vision, they need plan­ning, sync­ing, hard­ware manip­u­la­tion, nav­i­ga­tion. So you can’t just build a robot with one dis­ci­pline. You need a lot of dis­ci­plines to come togeth­er to real­ize an opti­mal solution.

There are many appli­ca­tions that we can think of for social robots. You can think of an exoskele­ton that knows where your friends are. You can think of drones that know the dif­fer­ence between civil­ians and sol­diers. A search and res­cue robot that knows when some­one is in pain. A car that knows that you’re dis­tract­ed. A house that knows that some­one does­n’t belong there. 

So while we are all think­ing about what robots can do, we should make robots think about us and what we do. So I’ve showed you that machines can respond and deal with social behav­iors, our social behav­iors, and I’m very inter­est­ed to know how social robot­ics could mean any­thing in your field inter­est or pas­sion. Thanks.