Facebook on Wednesday told its F8 conference audience about two new cutting-edge projects that could change the way humans engage with devices.
Over the next two years, the company will work on a new technology that will allow anyone to type around 100 words per minute -- not with fingers, but using a process that would decode neural activity devoted to speech.
What Facebook envisions is a technology that would resemble a neural network, allowing users to share thoughts the way they share photos today.
This technology also could function as a speech prosthetic for people with communication disorders or as a new way to engage in an augmented reality environment, suggested Regina Dugan, vice president of engineering at Facebook's Building8.
The other project announced at F8 would change the way users experience communication input -- that is, it would allow them to "hear" through the skin. The human body has, on average, about two square meters of skin that is packed with sensors. New technologies could use them to enable individuals to receive information via a "haptic vocabulary," Facebook said.
Putting the Best Interface Forward
About 60 employees currently are working on the Building8 projects -- among them machine learning and neural prosthetic experts. Facebook likely will expand this team, adding experts with experience in brain-computer interfaces and neural imaging.
The plan is to develop noninvasive sensors that can measure brain activity and decode the signals that are associated with language in real time.
The technology would have advance considerably before individuals would be able to share pure thoughts or feelings, but the current effort can be viewed as the first step toward that goal, said Facebook CEO Mark Zuckerberg.
Type With Your Brain
Facebook is not the only company working on technology that could allow a direct brain-computer connection. Elon Musk last month launched Neuralink, a startup dedicated to developing a "neural lace" technology that could allow the implanting of small electrodes into the brain.
One difference between Facebook's "type with your brain" concept and Musk's neural mesh project is that "Facebook wants it to be noninvasive," said futurist Michael Rogers.
"That's a tough one, because the electrical impulses in the brain are very small and the skull is not a great conductor," he told TechNewsWorld.
The nerve signals outside the skull that trigger muscles are much stronger, so reading brain waves noninvasively means filtering a lot of much-stronger noise.
"The existing 'neural cap' technologies that allegedly let wearers, say, learn to control computer games with their brains, are actually probably training them to -- unconsciously -- use their eyebrow and forehead muscles," suggested Rogers. "Interesting, but not the same as typing with your brain."
Hearing Through Touch
Just as Braille has allowed the blind to read, Facebook's technology to hear through the skin could be a game changer for those who are deaf.
"Sound through the skin sounds more like a tool for the profoundly deaf than for everyday use," observed Rogers -- and it might not be the best solution for those with such hearing problems.
"I'm still a big believer in the potential for bone conduction temples on smart glasses as a really good audio solution that doesn't require earbuds," he added.
Delivering information through this technology may be less challenging than processing it.
"It is technically easy to translate the data to a speech pattern, but the hard part is training the human to understand it," said Paul Teich, principal analyst at Tirias Research.
"Facebook can simplify the words to codes, but you are still stuck with the training to understand the code," he told TechNewsWorld.
"There is no way to immediately understand what is being sent, and this isn't very intuitive as I see it," added Teich. "With the right training, however, a coded message could be understood in real time."
Facebook's Timeline
Zuckerberg emphasized that Facebook is taking the first steps toward development of these technologies, and it could be a lot time before they have practical applications.
"Having bold visions and making ambitious predictions characterizes some of today's most regarded tech entrepreneurs," said Pascal Kaufmann, CEO of Starmind.
"There seem to be no limits, and one can easily fall prey to the belief that everything is possible," he said.
"However, climbing a high tree is not the first step to the moon -- actually, it is the end of the journey," Kaufman told TechNewsWorld.
"Despite some gradual improvements in speech recognition these days, understanding of context and meaning is still far beyond our technological capabilities. Taking a shortcut through directly interfacing human brains, and circumventing the highly complex translation from nerve signals into speech and back from speech recognition into nerve signals, I consider one of the more creative contributions in the last few months," he said.
"It is certainly an alternative to the brute force approaches and the unjustified AI hype that seems like climbing a tree rather than building a space rocket. Zuckerberg's announcement describes a space rocket; it is up to us now to develop the technology to aim for the moon," said Kaufman.
"Both of these technologies are well within reach. This kind of brain imaging has been used in academic settings for years, and scientists developed versions of 'skin hearing' devices over 30 years ago, " said Michael Merzenich, designer of BrainHQ, professor emeritus in the UCSF School of Medicine, and winner of the 2016 Kavli Prize.
"The real challenge is making this science practical. How can a brain be trained to learn to control these new devices?" he asked.
"Every new form of communication -- from writing to the telephone to the Web -- changes human culture and human brains," Merzenich told TechNewsWorld. "What impact will using these new devices have on our brains -- and our culture?" 
News , Technology
No comments:
Post a Comment