Slashdot Mirror
'Brain-to-Text' Interface Types Thoughts of Epileptic Patients
Jason Koebler writes with a link to Motherboard's article about research from the Schalk Lab of Albany, New York, where researchers "have just demonstrated for the first time that it's possible to turn a person's thoughts into a legible phrase using what they're calling a "brain-to-text" interface," writing "It's still still the early days of this technology—electrodes had to be placed directly on the brain and the 'dictionary' of phrases was limited. Still, brainwaves of thought patterns were turned into text at a rate much better than chance."
"That twelve year old girl sure is hot" Ummm, ehhh, seem to have a problem here...
Yes, I would like a happy ending with that.
This is my opinion based on what little I know and understand of the rumors and lies Thanks, Randal
According to TFA they could hook this up to anybody. It was just that epilepsy patients were having invasive surgery done anyway.
http://michaelsmith.id.au
Yeah, and I built a 'speech recognition' system in 1k when i was a kid. It recognised words, but to compare it to what we call 'speech recognition' is a joke. Sounds like these guys are doing someething pretty damn cool, so to have it sensationalised just hurts their cause and makes science in general look dumb.
No. it won't write a letter to mom from your thoughts, which is what the headline suggests. No i didn't RTFA.
This is no troll. This is a danger of this technology.
"Ow my brain."
systemd is Roko's Basilisk.
Anyone who had done this earns the right to dance around Kevin Warwick singing 'I'm more cyborg than you are' before ordering him to stop showing off to the public and get back to doing proper science.
https://en.wikipedia.org/wiki/...
Liberty - Security - Laziness - Pick any two.
... it says "transfer ... funds ... to ... doctor's ... account ...." - yep, that's what it says!
... why would I want to type thoughts of epileptic patients?
"Brainstorm" http://en.wikipedia.org/wiki/B...
"Win treats sysadmins better than users. Mac treats users better than sysadmins. Linux treats everyone like sysadmins."
<Seizure> OMGWTFBBQ. Urg. LOL. IDK? FU M8! Wut?
This is no troll. This is a danger of this technology.
It is both a danger and a promise of the technology. It can be used for brain-to-text, but also for other "mind-reading" tasks. A criminal suspect can be shown photos of the crime scene, or photos of the victim or other people known to be involved, and it is likely that it can detect whether the suspect recognizes the scene or person. This could be used to exonerate innocent people, but it could also be used as a tool of repressive authoritarianism to detect thought-crime.
Once this tech is fully matured, human sentience will be as scalable as the cloud!
The metabrain comes. I CAN'T WAIT!!!
Really, no. No more than an ordinary lie detector.
I used to design neural prostheses technologies. These direct neural control systems are limited partly by the size of the electrodes, which need to be stable without causing tissue damage, which limits their minimum size. They *cannot* be connected directly to individual neurons, or they will damage and disable the individual neurons over time. But since their minimum size is so limited, they are always, *always* sampling nerve signals from dozens if not hundreds, even thousands of neurons. And the electrical noise of embedding neurons in or next to normal nerve tissue is very high indeed: you pick up all the non-digital, non-clocked impulses generated signals from all the other neurons nearby. And these aren't normal analog voltage level signals: they're impulses that auto-generate and there is *always* surrounding electrical noise unless the tissue is quite dead or disabled.
The result is that, to detect a "significant" signal, the user needs to focus a quite consistent nerve signal for an extended period to overwhelm the background noise. So we're talking half a second per bit, based on the old myo-electric work for the Boston Arm, and similar work ever since. Until and unless someone creates a much more sophisticated neural interface, the bandwidth is far, far too low for "mind reading" or even for useful motor control. Some folks have tried to create better interfaces: David Edell over at https://www.linkedin.com/pub/david-edell/13/1bb/200 was transecting rat nerves and putting a silicon chip with gold plated holes across cut rat nerves, and showed that the nerves could grow back through the holes. But I'm not sure how far he's gotten.
And you do *not* want to transect brain tissue and plant electrodes into the cut parts, hoping to read minds as the nerves regrow through the silicon!
, partly by the
This is truly incredible what they have accomplished! I was sure when I read the headline that this was going to be another typical case of modern day Slashdot headline madness. I was shocked when I read both of.the attached articles to discover this really does sound like the first step in creating brain to computer interfaces that used imagined speech to control. Very, very cool. Rather surprised at the lack of comments on this one. Maybe everyone assumed it was BS and didn't even bother clicking?
"Burma shave"
-