AlterEgo, the device that turns thoughts into Google searches.
The MIT Media Lab at work on an instrument that perceives the ‘words thought’ thanks to the signals transmitted by the brain to the facial muscles. Researchers: “It’s like having superpowers”.
We are not in the vicinity of Cerebro, the imaginary technology used by Professor Charles Xavier in the X-Men comics and films to telepathically locate the other mutants scattered here and there across the globe thanks to a sort of global neural network. . Let’s talk about AlterEgo, the tool born at the MIT Media Lab by Arnav Kapur and Pattie Maes: it allows the wearer to perform searches on Google or calculations without saying a word.
· The AlterEgo Prototype
A hook that is still in the right ear and a band with sensors placed in seven key areas of the cheek, jaw and chin, the prototype can recognize the numbers from 0 to 9 and about a hundred words. According to its creators, instead of saying the usual “Hey Siri!” or “Ok Google” and then dictating to the virtual assistant the search for a sushi restaurant or the evolution of the weather situation on a particular geographic area, just “think” our query (the request). The outcome will then be communicated via the appropriate headset.
“It’s like having superpowers,” said Arnav Kapur even though, in reality, the explanation is much more prosaic: when we think of speaking, our brain sends signals to the facial muscles, even if we do not say anything aloud. AlterEgo is able to translate them.
“The motivation behind our research was to give life to an ‘augmented intelligence’ device.We worked around a very specific concept: it is possible to create a data processing system that is more internal and able to merge the human element and the artificial one so as to obtain a sort of internal extension of our own cognition? “. The answer – although it is still a project under development – seems to be affirmative: AlterEgo dialogues with the user and Google without any third party to be able to realize the exchange; the same earphone with which the answers are provided to the user does not totally shield the hearing, allowing a more immediate interaction with the surrounding environment.
In the research phases that led to the creation of the prototype, it was fundamental to understand what part of the face from which the most reliable neuromuscular signals are generated. To find out, they asked some ‘guinea pigs’ to sub-repeat the same series of words four times, while the movements were recorded by 16 electrodes that, at each sub-vocal repeat, were moved to a different area of the face: the information obtained in this session have led to the identification of the seven particular areas with which to communicate with AlterEgo.
· VOCAL RECOGNITION: 90%
The experiments of Kapur and Maes, at the time, have been done by giving AlterEgo questions focused on a predefined set of cities – such as “How many are the inhabitants of Santiago de Chile?” – And basic arithmetic problems. The ‘internal’ voice recognition required about 31 hours of training by the testers: the accuracy of the numbers and words is currently 90%. Both are convinced that the accuracy of AlterEgo can only increase with use.
This is not the only device of this kind to be under development, but it is certainly less ‘invasive’ for the convenience of use and the aesthetics of design than medical ones. The dr. James M. Gilbert of the University of Hull is working on a device designed for cancer patients who have lost the larynx, but its functioning revolves on the placement of magnets implanted on the lips and on the tongue: very difficult to emerge from the context of hospital employment. Still in the field of medical devices, it is difficult not to think of ACAT, the predictive interface developed by Intel that allowed Stephen Hawking to formulate sentences, then pronounced by the speech synthesizer, with a software keyboard able to perceive through the sensor to infrared positioned on the glasses the movements of the cheeks of the recently vanished luminary.
The question becomes more intricate with Facebook on one side and Elon Musk with his Neuralink on the other. Both realities are trying to build computer – brain interfaces capable of transforming thoughts into text by intercepting brain signals and not nervous ones.
Here then, the doubt that comes from thinking of a technology that takes information directly from human thinking, in which case the issues of privacy could take on even more disturbing features of the necessary reflections and mea culpa arising after the Cambridge Analytica scandal.