Man or machine?
MIT's humanoid robots showcase both human creativity and contemporary pessimism.
Humanoid robots were once the stuff of political and science fiction. Today, scientists working in Japan and the USA have been turning fiction into a physical reality.
During July 2003, the Museum of Science in Cambridge, Massachusetts exhibited what Honda calls ‘the world’s most advanced humanoid robot’, ASIMO (the Advanced Step in Innovative Mobility) (1). Honda’s brainchild is on tour in North America and delighting audiences whereever it goes. After 17 years in the making, ASIMO stands at four feet tall, weighs around 115 pounds and looks like a child in an astronaut’s suit. Though it is difficult to see ASIMO’s face at a distance, on closer inspection it has a smile and two large ‘eyes’ that conceal cameras (2).
The robot cannot work autonomously – its actions are ‘remote controlled’ by scientists through the computer in its backpack. Yet watching ASMIO perform at a show in Massachusetts it seemed uncannily human. The audience cheered as ASIMO walked forwards and backwards, side to side and up and downstairs. After the show, a number of people told me that they would like robots to play more of a role in daily life – one even said that the robot would be like ‘another person’.
ASIMO can punch the air (as if to say, ‘Yes!’), mimic shrugging its shoulders, clap, shake hands and take a bow. It can even dance to the Hawaiian Hulla (a scientist at the Massachusetts Institute of Technology (MIT) told me that robots often entertain Japanese executives at business meetings). These gesture signals are good to entertain the audience – but they are also part of a broader programme by scientists to facilitate robots’ entry into human culture.
Such robots say much about the way in which we view humanity, and they bring out the best and worst of us.
On one hand, these developments express human creativity – our ability to invent, experiment, and to extend our control over the world. On the other hand, the aim to create a robot like a human being is spurred on by dehumanised ideas – by the sense that human companionship can be substituted by machines (3); that humans lose their humanity when they interact with technology (4); or that we are little more than surface and ritual behaviours, that can be simulated with metal and electrical circuits.
While the Japanese have made huge strides in solving some of the engineering problems of human kinetics and bipedal movements, for the past 10 years scientists at MIT’s former Artificial Intelligence (AI) lab (5) (recently renamed the Computer Science and Artificial Intelligence Laboratory, CSAIL (6)) have been making robots that can behave like humans and interact with humans. For the past few months, I have been observing the day-to-day work of some of these MIT scientists.
One of MIT’s robots, Kismet, is an anthropomorphic head and has two eyes (complete with eyelids), ears, a mouth, and eyebrows. It has several facial expressions, including happy, sad, frightened and disgusted. Human interlocutors are able to read some of the robot’s facial expressions, and often change their behaviour towards the machine as a result – for example, playing with it when it appears ‘sad’. Kismet is now in MIT’s museum, but the ideas developed here continue to be explored in new robots (7).
Cog (short for Cognition) is another pioneering project from MIT’s former AI lab. Cog has a head, eyes, two arms, hands and a torso – and its proportions were originally measured from the body of a researcher in the lab. The work on Cog has been used to test theories of embodiment and developmental robotics, particularly getting a robot to develop intelligence by responding to its environment via sensors, and to learn through these types of interactions (8).
This approach to AI was thought up and developed by a team of students and researchers led by the head of MIT’s former AI lab, Rodney Brooks (now head of CSAIL), and represented a completely new development (9). Another radical approach developed by Brooks was behaviour-based robotics. While conventional approaches to AI robots might want to map all the elements that are in the robot’s environment so that its behaviour could be completely pre-planned beforehand, behaviour-based robotics focuses instead on getting the machine to act autonomously.
This work at MIT is getting furthest down the road to creating human-like and interactive robots. Some scientists argue that ASIMO is a great engineering feat but not an intelligent machine – because it is unable to interact autonomously with unpredictabilities in its environment in meaningful ways, and learn from experience. Robots like Cog and Kismet and new robots at MIT’s CSAIL and media lab, however, are beginning to do this.
These are exciting developments. Creating a machine that can walk, make gestures and learn from its environment is an amazing achievement. And watch this space: these achievements are likely rapidly to be improved upon. Humanoid robots could have a plethora of uses in society, helping to free people from everyday tasks. In Japan, for example, there is an aim to create robots that can do the tasks similar to an average human, and also act in more sophisticated situations as firefighters, astronauts or medical assistants to the elderly in the workplace and in homes – partly in order to counterbalance the effects of an ageing population.
Alternatively, and more pessimistically, researchers are exploring the possibilities of building robots to act as companions to humans or to spy for wives on their businessmen husbands (10). A Honda representative quipped that the head of Honda wanted an ASIMO in every factory to keep watch on the employees. Robotic dogs like AIBO (produced by Sony) are proving to be exceptionally popular as companions in the densely populated city of Tokyo (11).
So in addition to these potentially creative plans there lies a certain dehumanisation. The idea that companions can be replaced with machines, for example, suggests a mechanical and degraded notion of human relationships.
Being human is about much more than merely performing correctly in response to particular signals. And the very idea of a mechanical companion is only conceivable in a cultural context where human beings feel a greater sense of atomisation from each other. The growth of the awe and faith placed in machines, rather than the people who make them, says more about how downgraded our own potential as human beings has become.
The tension between the dehumanised and creative aspects of robots has long been explored in culture. In Karel Capek’s Rossum’s Universal Robots, a 1921 play in which the term ‘robot’ was first coined, the robots resemble and behave like humans, and were made of flesh and blood; in the play, they were actually mistaken for human beings (12). In Fritz Lang’s film Metropolis (1926), the character Rotwang creates the robot Maria, which is first made out of metal, and later transformed into the appearance of a human being (13).
But although both Capek’s and Lang’s robots had human-like appearance and behaviour, neither dramatist thought these robots were human. For Capek and Lang, being human was about much more than appearing to be human. In part, it was about challenging a dehumanising system, and struggling to become recognised and given the dignity of more than a machine.
A similar spirit would guide us well through twenty-first century experiments in robotics.
Kathleen Richardson is a social anthropologist studying for a PhD at the University of Cambridge. She is currently conducting her fieldwork in the humanoid robotics lab at the Computer Science and Artificial Intelligence Lab CSAIL at the Massachusetts Institute of Technology (MIT).
(1) Details of the ASIMO tour in Cambridge, MA
(2) Honda’s Website for ASIMO
(3) Robots as Companions/Sociable Machines at the Media Lab
(4) Donna Haraway’s Cyborg Manifesto
(5) AI website
(7) Kismet, Humanoid Robotics Group, MIT Artificial Intelligence Laboratory
(8) Humanoid Robotics Group, MIT Artificial Intelligence Laboratory
(9) For some explanation of behaviour based robotics see Flesh and Machines: How Robots Will Change Us, by Rodney Brooks, Pantheon Books, 2002. (buy this from Amazon UK or Amazon USA).
(10) A book about scientists who make robots and what they think of them: Menzel P and Faith D’Alusio, Robosapiens: Evolution of a New Species, MIT Press, 2000 (buy this from Amazon UK or Amazon USA).
(11) AIBO website, Sony
(12) R.U.R. and The Insect Play, by the Brothers Capek
(13) Metropolis – outlining the story and details of the film
To enquire about republishing spiked’s content, a right to reply or to request a correction, please contact the managing editor, Viv Regan.