麻豆传媒

In a room surrounded by machines at the Massachusetts Institute of Technology鈥檚 Media Lab, a group of students and professors is sharing a deeply human moment. On a video screen Amanda Baggs, a 26-year-old woman with autism, is flapping her arms, rocking, sucking on a pen and scratching household objects, part of an lifted from video-sharing website YouTube.

鈥淔ar from being purposeless, the way that I move is an ongoing response to what is around me,鈥� says the robotic voice that narrates her typed words. The class reacts emotionally 鈥� not out of pity but in response to Baggs鈥檚 proud and moving appeal for viewers to recognise 鈥渢he existence and value of many different kinds of thinking鈥�.

It is precisely the value of Baggs鈥檚 kind of thinking that has drawn this novel class together. By combining research into autism with the technology of sociable computing, which seeks to endow computers with social skills, organisers hope to develop a new approach to both fields.

鈥淎utism is of great interest to me,鈥� says Rosalind Picard, the sociable computing researcher at the Media Lab who kicked off the class鈥檚 inaugural session last week. 鈥淚t turns out we are trying to solve a lot of the same problems that people who study autism are trying to solve.鈥�

For years, Picard has described machines as autistic because of their inability to empathise, interpret facial expressions and generalise from one situation to another. She has pioneered technologies that sense emotions, including StartleCam, a wearable video camera that saves footage only when the skin conductance of the wearer suggests excitement. Now, she says, it is time to turn these 鈥渕ind-reading鈥� technologies into tools for people with autistic spectrum disorders (ASD), which, according to newly released figures affect 1 in 150 US children. At the same time, says Picard, the way people with ASD interact with the world can inform the design of more sociable robots. Teaching the class with her are Cynthia Breazeal, widely known in robotics circles for creating the emotional robot Kismet, and Sherry Turkle, who studies the relationships people form with technology.

鈥淚t鈥檚 time to turn 鈥榤ind-reading鈥� technologies into tools for autistic people鈥�

Although the researchers do not seek to 鈥渃ure鈥� ASD, which they see not as a disability but as an alternative, sometimes advantageous, mental state, tools that provide autistic people with a 鈥減rint-out鈥� of other people鈥檚 emotions could serve as useful aids to interaction. 鈥淥ften people with autism want social contact but can鈥檛 get it,鈥� says post-doctoral student Rana el Kaliouby.

Turkle, who in the past has criticised the practice of using robots as companions for the elderly, feels differently about devices designed to enhance interactions for people with ASD because the electronic tools do not replace humans, but rather do something they cannot. 鈥淭his is the first case where a robot has a competitive edge,鈥� she says.

But the class is as much about building more capable robots and software programs as it is about helping people. Because people with ASD tend to be highly systematic, the way they approach social interactions might be easier to recreate in a computer program than the way a 鈥渘eurotypical鈥� person does, says el Kaliouby. 鈥淩obots are going to need to know when they can interrupt people, or people are not going to buy them,鈥� says Picard. Participants acknowledge the link between ASD and robot communications should not be taken too literally, or it risks putting people with autism on a par with machines. 鈥淚t鈥檚 important to remember this is a metaphor,鈥� says Rebecca Waber, a graduate student taking the class who has an autistic sibling. On the screen, Amanda Baggs鈥檚 unmistakably human statement of identity seems to echo the sentiment.

鈥淭he class is as much about building capable robots as it is about helping people鈥�