The MIT Computer Science and Artificial Intelligence Laboratory (CSAIL) has been working with Boston University experts on a system to enable robots to correct their own errors by interpreting signals from a human brain. Does this herald the start of real dialogue between Man and Machine?
“When you talk about Man you are talking about language, and when you are talking about language you are talking about society,” argued French anthropologist and philosopher Claude Levi-Strauss, and today language is playing an even more central role in our cultural journey. In a world of hyper-communication, even linguistic tools can have a disruptive effect. Today computer code is adding to the overall potential of language and is thus changing our relationship with the world. Lines of code form a vital bridge between the way we communicate and our new digital tools, and are nowadays a major driver of innovation and creation.
To date it has always been people who directly governed all interactions with our help-mates, the machines. However, with recent progress in robotics and artificial intelligence (AI), a new channel for Man-Machine communication is emerging – pure thought. And after all, what could be more logical than to enable intelligence – both human and artificial – to interact through the medium that most characterises intelligence? If establishing real dialogue between people and robots still seems to belong to the realms of fantasy, the work being done by the teams at CSAIL and Boston University is nevertheless raising hope that one day we will be able to develop genuine communication with machines, just by harnessing the power of brain waves.
At the core of the experiment is a system that enables a team member to correct errors made by Baxter, a robot built by Boston-based Rethink Robotics, solely through thought. Sensors placed on the head of the operator send brain signals via a feedback system to the robot. The system decrypts the signals and analyses them, then delivers a simple command to Baxter. If this goes forward, one day there will no longer be any need for a mechanical interface between people and robots. Messages will be conveyed through thought waves, a much less demanding and time-consuming process.
The tests carried out so far by the CSAIL/Boston University team indicate real progress. The team’s ground-breaking machine-learning algorithms enable the system to classify brain waves in the space of 10 - 30 milliseconds, which means that the probability of error in a given action can be calculated very fast, enabling Baxter to re-orient his actions accordingly. This is one way of increasing a robot’s capacity to act independently, but not least it may represent a new form of collaboration between Man and Machine.
In the longer term, human thought could be employed as a sort of remote control connected to its robotic/digital environment. This would open up the potential for innovation in a number of fields of work, healthcare and the smart home. But the ultimate aim of the CSAIL/Boston University research is to make the robot of the future a sort of extension of the human brain. It is certainly not unthinkable that as the new generations of robot become even more sophisticated, they might be equipped to receive a multitude of different brain signals, implying a whole range of commands and input skillsets. At that point, it might be legitimate to talk about the beginnings of real Man-Machine dialogue, about continual spontaneous communication, about collaboration between mind and technology.
There is however one aspect that cannot be ignored: this kind of dialogue is basically utilitarian and operational. And it only has one voice and one will – i.e. one-way commands from the human master, meeting with obedience on the part of the machine. Still, while French intellectual and literary figure Georges Bataille argues that: “We would no longer have any human qualities if our speech were totally subservient”, the reverse is probably true for robots.