The officials of Facebook announced that they have the plan to make an artificial intelligence based device that will allow the user to type those words just by thinking. The firm funded a study for developing a machine-learning algorithm that is capable of switching the brain activity into speech.
The working of the device depends on epilepsy patients who are already recording electrodes put on their brains to evaluate the origins of their seizures, ahead of brain surgery. Whereas, the social media giant is hoping that it will pave the path for a completely non-offensive, wearable device that has the ability to process a hundred words per minute.
Muscle Twitches of Patients
The scientists of the University of California San Francisco quizzed a list of multiple-choice questions that ordered randomly and requested them to answer loudly.
And the algorithms learned to recognize:
- the asked question, 75% of the whole time
- the picked answer of patients, 61% of the entire time
Prof Eddie Chang said that most recent tactics focused on decoding speech only, but here they represent the worth of decoding both angles of a conversation – both the questions anyone hears and what they say in reply.
It gives fresh blood to their perception that speaking is not something that happens in a vacuum and that any effort to decode whatever patients with speech hurts are making efforts to say will improve with speaking into account the whole context in which they are attempting to communicate.
He added, at present patients that are unable to talk due to paralysis are partial to spell out words very slowly with the help of muscle twitches or residual eye movements to control a computer interface.
Whereas, in several cases, info required to make smooth speech still stuck in their brains. And they needed technology or any AI-based device that allows them to express it.
Future Plans to Increase the Efficiency
David Moses, the researcher, said that they reached this level with the help of limited vocabulary whereas in future researches have positive faith that they will increase the efficiency along with the accuracy of what they will able to translate directly from the brain activity.
Furthermore, Facebook elaborated in its blog that finally, the researchers have hope to attain an actual decoding speed of hundred words per minute having a one thousand word vocabulary where the rate of word error is less than 17%.
Either than looking towards the screen of mobile phone or laptop, they can keep eye contact and retrieve valuable info and perspective without losing a single beat. For the moment, Neuralink organization of Elon Musk applied to the United States regulators for getting a permission letter to test its brain-hacking device on humans.