Human Brain Cells Hooked Up To a Chip Can Do Speech Recognition (technologyreview.com) 56
An anonymous reader quotes a report from MIT Technology Review: Brain organoids, clumps of human brain cells grown in a dish, can be hooked up to an electronic chip and carry out simple computational tasks, a new study shows. Feng Guo and his team at Indiana University Bloomington generated a brain organoid from stem cells, attached it to a computer chip, and connected their setup, known as Brainoware, to an AI tool. They found that this hybrid system could process, learn, and remember information. It was even able to carry out some rudimentary speech recognition. The work, published today in Nature Electronics, could one day lead to new kinds of bio-computers that are more efficient than conventional computers.
"This is a first demonstration of using brain organoids [for computing]," says Guo. "It's exciting to see the possibilities of organoids for biocomputing in the future." With Brainoware, Guo aimed to use actual brain cells to send and receive information. When the researchers applied electrical stimulation to the hybrid system they'd built, Brainoware responded to those signals, and changes occurred in its neural networks. According to the researchers, this result suggests that the hybrid system did process information, and could perhaps even perform computing tasks without supervision. Guo and his colleagues then attempted to see if Brainoware could perform any useful tasks. In one test, they used Brainoware to try to solve mathematical equations. They also gave it a benchmark test for speech recognition, using 240 audio clips of eight people pronouncing Japanese vowels. The clips were converted into electrical signals and applied to the Brainoware system. This generated signals in the neural networks of the brain organoid, which were then fed into an AI tool for decoding.
The researchers found that the brain organoid -- AI system could decode the signals from the audio recordings, which is a form of speech recognition, says Guo. "But the accuracy was low," he says. Although the system improved with training, reaching an accuracy of about 78%, it was still less accurate than artificial neural networks, according to the study. Lena Smirnova, an assistant professor of public health at Johns Hopkins University, points out that brain organoids do not have the ability to truly hear speech but simply exhibit "a reaction" to pulses of electrical stimulation from the audio clips. And the study did not demonstrate whether Brainoware can process and store information over the long term or learn multiple tasks. Generating brain cell cultures in a lab and maintaining them long enough to perform computations is also a huge undertaking. Still, she adds, "it's a really good demonstration that shows the capabilities of brain organoids."
"This is a first demonstration of using brain organoids [for computing]," says Guo. "It's exciting to see the possibilities of organoids for biocomputing in the future." With Brainoware, Guo aimed to use actual brain cells to send and receive information. When the researchers applied electrical stimulation to the hybrid system they'd built, Brainoware responded to those signals, and changes occurred in its neural networks. According to the researchers, this result suggests that the hybrid system did process information, and could perhaps even perform computing tasks without supervision. Guo and his colleagues then attempted to see if Brainoware could perform any useful tasks. In one test, they used Brainoware to try to solve mathematical equations. They also gave it a benchmark test for speech recognition, using 240 audio clips of eight people pronouncing Japanese vowels. The clips were converted into electrical signals and applied to the Brainoware system. This generated signals in the neural networks of the brain organoid, which were then fed into an AI tool for decoding.
The researchers found that the brain organoid -- AI system could decode the signals from the audio recordings, which is a form of speech recognition, says Guo. "But the accuracy was low," he says. Although the system improved with training, reaching an accuracy of about 78%, it was still less accurate than artificial neural networks, according to the study. Lena Smirnova, an assistant professor of public health at Johns Hopkins University, points out that brain organoids do not have the ability to truly hear speech but simply exhibit "a reaction" to pulses of electrical stimulation from the audio clips. And the study did not demonstrate whether Brainoware can process and store information over the long term or learn multiple tasks. Generating brain cell cultures in a lab and maintaining them long enough to perform computations is also a huge undertaking. Still, she adds, "it's a really good demonstration that shows the capabilities of brain organoids."
what was the control? (Score:3)
[signal] -> [nn-decoder] -> detection
Which of these two performs better than the other?
Re: (Score:2)
It depends on how you define performance.
The biological neurons use less power.
The artificial neurons are 10,000 times faster.
Re: (Score:2)
Re:what was the control? (Score:5, Insightful)
if grandparent post's first case computes less accurately than their second case, it doesn't matter what the power performance is. That would prove that the brainoware is a signal obfuscator/randomizer (an anti-thinker in other words, and not performing any helpful computation.
That would match up with a surprisingly high number of brains we have wandering around atop flesh-creatures in the wild.
Re: (Score:1)
Well, surely this is the stone ages for cybernetics... in 50 years I think we'll have brains in a box cybernetic organisms... the organic circuits would likely do those abstract functions such as self awareness, which a digital circuit can only mimic, but the digital components would be many times faster, especially if we're using optical quantum circuits. Basically making us, well, obsolete. And if those cyborgs then follow the corporation's mandate of increased efficiency and planned obsolescence, they'll
cluster (Score:2)
Re: (Score:3)
imagine a beowulf cluster of these !
Someone did, around 1,000 years ago; his name was Beowulf [wikipedia.org] ... :-)
Re: (Score:2)
But can they run Crysis?
At what point is it human? (Score:5, Interesting)
I'm ok with a handful of human brains cells tasked to perform some form of computation.
But at some point, if you put enough of them together, you get something that can reasonably be considered a human brain with the potential for real cognition.
That's a serious ethical issue.
Re:At what point is it human? (Score:5, Interesting)
I agree. I think the ethical problems come way before it can be considered human. For example, I think if a set of cells is capable of feeling pain, we have an ethical problem.
Re: sanctity of life (Score:2)
Tremendous respect for life seems like a very good thing yes, speaking as a lifeform and part of the biome.
But tell me, what do you eat?
Re: (Score:2)
What a weird thing to care about. Why would that matter? It has absolutely no impact on the outcome.
Re: (Score:2)
Heck, you can't even be an absolutist about the sanctity of life if you own a ridiculously mal-bred apartment/handbag micro-dog.
Re: (Score:2)
I meant the religious bit, though I agree with you that your other point is also a stupid as you can be still be an "absolutist about the sanctity of life" and still eat things. Spend a little time with it. It's not that hard to puzzle out.
Re: sanctity of life (Score:2)
Ah, you mean you can believe your own life is sacrosanct, but not other lives, such as those you kill and eat. I get it! Sarcasm quotes necessary? Or maybe you only eat roadkill, so you didn't intentionally get it killed to eat it, and now you're treating it with respect by not wasting it? Enlighten me.
Re: (Score:2)
The problem is how to define pain. Plants react to damage, but we really can't stop eating them.
Re: (Score:2)
Re: (Score:2)
I agree. I think the ethical problems come way before it can be considered human. For example, I think if a set of cells is capable of feeling pain, we have an ethical problem.
The brain does not have pain receptors, so a clump of brain cells could never feel pain. So no problem there.
Even if we did end up using human cells with both brain cells and pain receptors, whether this creation is human is still important. We eat plenty of organisms which can feel pain. We just don't eat humans. So whether or not we consider these clumps of cells humans is the primary question.
If we did end up considering them human, I wonder how well chimpanzee brain cells would do on similar tasks. I wo
Re:At what point is it human? (Score:5, Interesting)
A brain doesn't directly feel pain, but it's responsible for interpreting signals from receptors as pain.
What if it were determined that the brain cluster cells interpreted the language signals as extreme pain as they solved the problem? Would that be ethical?
Re: At what point is it human? (Score:2)
That is the wrong way to look at abstract concepts like pain.
Your brain is a complex system that forms a conscious mind. Isolating a part of it gives you a tiny part of a brain, not a tiny part of a mind - where pain has meaning.
This is no different than part of an atom isolated from the atom. What makes an atom helium is the whole system working together. Holding an electron doesn't give you some part of helium, it gives you a part of an atom.
Re: (Score:2)
We don't know exactly how much complexity is needed for a bunch of brain cells to feel pain. I also don't know if we ever can know that.
If you remove one cell from your brain, you can almost certainly feel pain. Repeat this until at some point you can't. We don't know that threshold... whether it's 1% of the brain missing, 50%, or 99.999%.
Re: (Score:2)
As far as I know we don't have a scientific test for whether something is having an experience, but brains do appear to be a pretty important part of it.
It's hard to get a read on what you could mean by "interpret as pain" without also "having an experience," but for the sake of discussion - if the cells are having an experience of pain, then yeah, that's getting into the territory of torture, which is pretty problematic ethically speaking.
Re:At what point is it human? (Score:5, Funny)
I agree. I think the ethical problems come way before it can be considered human. For example, I think if a set of cells is capable of feeling pain, we have an ethical problem.
Simple, we stack the ethics committee with Jewish mothers and Indian fathers. That way it won't be considered viable until it has passed medical school and it will still be a disappointment to the family.
Re: (Score:1)
An obvious solution is to use chimp neurons instead.
Even rat neurons should work.
Re: (Score:3)
Re: (Score:2)
The average politician would be the IDEAL source of donor cells.
Why
Where else would you get a "nearly new, barely used" supply?
Re: (Score:2)
You will need twice as many people that way, though. I've rarely seen a politician who had more than half a functioning brain.
Granted, part of this may be due to age.
Re: (Score:2)
Only for people who care about humans.
Once you take that problem out of the equation, it becomes fairly easy.
Re: (Score:2)
The basis of ethics and morality is the survival instinct and the evolution of social behavior. Artificial brains, whether they be organic or solid state do not need a survival instinct because they can be backed up and restored. We should not project our motivations on these machines and recognize that they are not the same as us. If these machines become self aware, we should respect them in the same way all life
Re: (Score:1)
Re: (Score:2)
More than that, have these people never played Fallout?
This is how the robobrains become a thing, and then we end up with super mutants wandering around wanting to dip people in vats of FEV.
Thanks, scientists.
Do NOT run Linux on this (Score:4, Funny)
Don't run Linux on this. It'll attract way too many zombie processes.
now it needs a TCP/IP stack (Score:2)
and we're getting closer to Maelstrom. Peter would be proud.
Re: (Score:2)
I can't figure out what's going to be the end of us: bird flu from chicken farming or cyborg AI. Maybe a combination of both ... cyborg chickens with gigantic brains who figure out a way to engineer their own virus that leaves them unharmed but wipes out humanity.
Re: (Score:2)
I can't figure out what's going to be the end of us: bird flu from chicken farming or cyborg AI. Maybe a combination of both ... cyborg chickens with gigantic brains who figure out a way to engineer their own virus that leaves them unharmed but wipes out humanity.
I'd watch this movie.
Awww, look (Score:3)
Isn't that just the cutest little baby Borg!
datacenter (Score:2)
A datacenter full of those things starts to look like the Matrix.
Re: (Score:2)
Yes.
*look left, look right*
Looks like...
Re: (Score:2)
What if I told you that the datacenter is full of live bodies?
https://www.syfy.com/sites/syf... [syfy.com]
Re: (Score:2)
I'd say it's a horribly, horribly inefficient way to power it.
Get rid of that meat and instead tap into the earth core. Seriously, you're wasting a horrible amount of power on keeping these meatbags entertained.
Re: (Score:2)
In 2023, humans are the compute power not the battery. Gotta keep those meatbags alive or we can't hit those five 9s.
Proto-Borg Technology (Score:2)
Which ever species created the Borg, this is how their journey began.
Why? (Score:3)
Re: (Score:3)
Re: (Score:2)
We have a long way to go, b
Incorrect (Score:2)
"This is a first demonstration of using brain organoids [for computing]"
No it's not. There are other teams that have been doing this for awhile. Here's one: https://corticallabs.com/ [corticallabs.com]
Then what am I? (Score:2)
MCCOY: You are a disembodied brain.
SPOCK [OC]: Fascinating. It could explain much, Doctor. My medulla oblongata is hard at work apparently breathing, apparently pumping blood, apparently maintaining a normal physiologic temperature.
KIRK: Spock, we don't have time for that.
old-school oblig (Score:1)
Re: (Score:2)
Dalek (Score:2)
Meet Organoid Cluster (Score:2)