- 4 months ago
A swiss company has succeeded in linking ‘mini-brains’ called organoids with microchips. Is this the first step toward building computers based not on silicon, but on biological material and ‘living’ neurons?
Category
🤖
TechTranscript
00:06A super computer made of human brain cells?
00:09Sounds like science fiction, but it's already becoming reality.
00:13A Swiss company is connecting microchips up to mini-brains, tiny clusters of cultured
00:19neurons known as organoids.
00:21Could we one day create an enhanced brain in the lab?
00:26Superbrains, AI and Space Exploration on DW's Science Show.
00:31Welcome to Tomorrow Today.
00:49You feel a sense of awe and humility knowing that this is a human
00:54brain, one from a person who donated their body to science.
00:59It was once the center of a human being.
01:02It controlled that person for very likely over 75 years.
01:06It was the seat of their memory, experiences, of sensations.
01:11It's where their heartbeat accelerated when they fell in love.
01:21This is the brain's frontal pole and this is the posterior pole.
01:26If we turn it, we can see the temporal lobe here on the side, which is set slightly apart
01:31from the rest.
01:33At the back, there's the cerebellum.
01:35It has no convolutions but instead is folded like leaves.
01:40Down below, you can see, between the two hemispheres, the brain stem and the midbrain.
01:46And here is where it transitions into the spinal cord.
01:5286 billion neurons make up the incredibly powerful thinking apparatus that is the human brain.
01:58But while we understand many aspects of how individual neurons work, despite decades of
02:04research, much remains a mystery, like how they give rise to memory, planning and consciousness.
02:12Think of it as a kind of company with different departments.
02:15Each department works somewhat independently and follows its own protocols.
02:19But for the company to function, they all have to work together.
02:22The same applies to the brain.
02:25Now, with this brain knife, which I have to moisten so it can glide easily through the
02:29tissue, I'll slice open the brain.
02:37Our brains may be the most complex structures in the universe.
02:42They have tremendous capabilities.
02:44Can a system this complex be recreated?
02:49Martin Kutter grows brain organoids, clumps of neurons that could one day become the computers
02:55of the future.
03:00What's incredible about this is that we can scale it.
03:03I can have a thousand neurons or if I want ten thousand, a hundred thousand, I just wait
03:07a little longer and I have them.
03:09And what will that bring?
03:10What's your vision?
03:12Superintelligence.
03:14Seriously?
03:15I don't know.
03:17Something that is super creative, super logical, with huge storage capacity.
03:24We don't know yet.
03:25And then we don't know yet.
03:30Kutter spent years researching conventional AI, only to see US researchers seize an insurmountable
03:37lead in the field.
03:39For the past decade, he's pursued a new approach, a kind of biocomputer.
03:45In it, skin cells are turned into stem cells, which are in turn used to grow brain cells.
03:55Biochemist Flora Brozzi has cultivated around 3,000 brain organoids so far.
04:01At first, they survived for just a few days.
04:04Now they live up to three months.
04:09Several of these neuron clusters, made up of about 1,000 brain cells, are placed directly
04:16onto a microchip.
04:20Then comes the crucial step.
04:23On the chip, the neurons multiply, differentiate, and connect with electronic contacts.
04:29The goal is to control the organoids via a computer.
04:36It grows and the cells connect and specialize.
04:40Different types of neurons emerge.
04:42And at some point, spontaneous activity begins.
04:45Maybe it thinks.
04:46I don't know what it thinks, but it's active.
04:49Does that make it something human for you?
04:52No, for us it's simply a material.
04:54These are neurons, cells, not something human.
04:58Kutter believes that in this form, the organoids will not develop consciousness
05:03because they lack a brain's complexity.
05:05They're just clusters of cells.
05:07With no blood vessels, the clumps currently only grow to be about 50,000 neurons in size.
05:14There are connections forming everywhere, and this is all spontaneous?
05:19Yes, as soon as they detect a new environment, they start connecting and communicating.
05:27If the vision turned into reality, biocomputers could have major advantages over conventional
05:33computers.
05:34They could be up to a million times more energy efficient, and they might be able to do something
05:40else, adapt.
05:45There are many hypotheses about what creativity actually is.
05:50Here, because there's spontaneity and constant activity, the structure of the organoid changes
05:55continuously, so new ideas could emerge even without prompting.
05:58And that's something silicon can't do.
06:08Now I can make a horizontal cut with the brain knife.
06:15There's some resistance.
06:16It's not that easy to cut through.
06:30The grey matter contains the nerve cell bodies, the neurons themselves.
06:35The white matter contains the connections between them.
06:38And those connections may be even more important than the cells themselves.
06:49In the brain, it's impossible to separate hardware and software.
06:54Whenever we see, hear, taste or understand something, it activates similar neural patterns.
07:02Yet at the same time, the brain rewires itself.
07:06In a way, with every thought, it becomes a different brain.
07:11The brain is constantly changing.
07:16That's what distinguishes it from machines or computers.
07:19It has a kind of flexibility that electronic systems don't have.
07:25Recreating this complexity artificially, without knowing exactly what input produces which function,
07:31that seems like a pretty difficult task.
07:39Our freshly raised brain organoids come into the incubator, which is set at 37 degrees Celsius.
07:46Like in the human body?
07:48Like in the body, 80% humidity, 5% CO2.
07:53It goes in here and is connected to the fluidic system.
07:56Nutrients are pumped in here and waste products are carried away.
08:06Could I control it now?
08:09We can do that.
08:10We can start a simulation, then you can simulate it and it'll respond.
08:16And will I understand it?
08:19If you did, you'd receive a Nobel Prize.
08:22You send a signal to the organoid, it's stimulated and it responds.
08:27Look, shortly after there's an answer.
08:30So, I'm communicating with this mini-brain in the cabinet?
08:34Exactly.
08:36And indeed, on close inspection, tiny changes in activity appear with every keystroke.
08:45Next, the goal is to program the organoids.
08:48The brain cells should rewire themselves all on their own,
08:52strengthening certain connections while weakening others,
08:55the same process that happens when we learn.
09:00For that, the organoid is rewarded with dopamine when its output,
09:05represented as spikes, looks the way the researchers want.
09:11If the output doesn't, the organoid is punished with electrical noise,
09:17meant to weaken the synapses.
09:22And does it work?
09:25It works reasonably well.
09:26We can store one bit of information and retrieve it.
09:31But the effect doesn't last very long, not even for days.
09:35One bit of information?
09:37So far.
09:38That's not very much.
09:40It's more than zero.
09:49Each neuron can essentially be seen as an analog computer.
09:55Like any analog computer, it has a central unit,
09:59the cell body containing the nucleus, where DNA is stored and proteins are produced.
10:06Then there's the dendritic tree,
10:08these antenna-like structures that extend outward.
10:12They form connections with other nerve fibers and receive information,
10:17which is processed locally in the cell body.
10:20A single neuron can integrate up to 100,000 inputs,
10:25from which it calculates a kind of result.
10:30And this result is then transmitted to other regions of the brain.
10:41Using a special high-tech microscope, scientists can visualize neural connections with a laser.
10:50Even a mouse brain has incredibly dense networks, including links between distant areas.
10:59The human brain is made up of around 900 million kilometers of neural wiring,
11:06with over 100 trillion synapses.
11:09But how signals are transmitted also depends to some extent on which neuromodulators have been released.
11:18This variability and specificity are so extraordinary that it's hard to compare the brain to any other known system or
11:26organ.
11:30This is a virtual environment, a software program running locally.
11:36It's connected to an organoid inside the incubator.
11:41Together they emulate the behavior of a moth.
11:47That means when the moth sees a light source, it asks the organoid, what should I do?
11:52And the organoid replies, move towards the light.
11:55So the organoids are essentially playing a kind of video game.
11:58Exactly.
12:00A fundamental biological behavior being recreated by a mini-brain?
12:04For moths, and for us too.
12:08Because it shows that the overall system works, we can connect biology with electronics and computer science, and it works.
12:17Conventional computers will never be replaced completely by brain organoids,
12:22but biocomputers could provide advantages when highly complex computations are necessary,
12:27like those involving artificial intelligence.
12:31When it comes to processing and storing information, creating new things, imagining things that don't yet exist,
12:38I believe this technology could make that possible.
12:44I'm not opposed to it as long as ethical and legal standards are maintained.
12:53I think it's good to try.
12:57Though I'm not sure I'll still feel that way if it actually works.
13:10Artificial intelligence still isn't as flexible or creative as the human mind.
13:15Instead, AI models learn from information that's already out there,
13:19including content from sources like Wikipedia.
13:22But the world's largest online encyclopedia is written by human volunteers.
13:28And humans sometimes make mistakes.
13:31So what happens when errors make their way in?
13:37Bärbel Mimitz has been active on Wikipedia for five years.
13:41The pensioner has written many articles, mainly on topics related to women.
13:46And every Tuesday she meets in person with other Wikipedians.
13:51I'd say it takes up at least 20 hours a week, writing, uploading, correcting, sometimes translating.
14:02This time, ten volunteers have come together in Hanover.
14:06They meet in one of the six Wikipedia offices in Germany,
14:10rented so the platform's authors and editors can collaborate and train.
14:16Contributors, however, have grown scarce.
14:18Fifteen years ago, around 11,000 were active.
14:22Now only 5,000 to 6,000 people regularly take the time.
14:27Today's session is about tips and tricks.
14:30Thousands of unreviewed articles are waiting to be edited.
14:33The volunteers can barely keep up.
14:37That led a major German newspaper to ask, how well does Wikipedia still work?
14:43Its analyzers randomly selected articles and found numerous errors.
14:53We were very surprised that it got the highest mountain in Sweden wrong.
14:58The former highest peak had a glacier that melted.
15:02Now another peak is higher, and that wasn't updated.
15:05That was the most surprising case.
15:08But there were also mistakes with political relevance.
15:12At the time of our study, the Fordo nuclear facility in Iran was in the news,
15:17the one that was bombed by the US.
15:20The Wikipedia article indicated that activities there could violate the Nuclear Non-Proliferation Treaty,
15:26which was incorrect.
15:28Details like these matter, especially in such situations.
15:35Mistakes were discovered in 198 German Wikipedia articles out of the 1,011 randomly selected for the study.
15:43Another 203 had outdated information.
15:48I expected some outdated content, but I didn't expect so many errors.
15:53Even in historical articles that basically can't become outdated.
15:58That surprised me.
16:02Among volunteer authors in Hanover who helped create this content, the analysis hit hard.
16:12Forty percent, that's quite serious, I have to say.
16:18As a community, we need to address it.
16:20It was quite a blow.
16:24In Berlin, at Wikimedia, the non-profit foundation that runs the platform but doesn't generate content,
16:32people have been trying for years to attract younger contributors.
16:35There are outreach programs in schools and universities, and new authors do join.
16:43They're all very committed, and many continue to stay involved.
16:52Personally, I wasn't surprised.
16:58Wikipedia is a volunteer project.
17:02So, of course, it's not 100% complete or correct.
17:07It'll never be finished.
17:09Wikipedia will never be finished.
17:12The open source project is the largest encyclopedia in human history.
17:17Here in Berlin, Wikimedia provides only the software and the servers to run it.
17:23But with a dwindling number of volunteers and a heightened awareness of mistakes, could the project fail?
17:30Wikimedia is trying to keep the platform easier for contributors to use, so that writing, editing, reviewing and publishing remain
17:38accessible.
17:39But the newspaper studies authors fear that's not enough.
17:46Our impression is that after about 20 years, volunteer work alone may not be enough to maintain the product's high
17:55quality.
17:58Some experts say that professional paid staff is now necessary.
18:05Paid contributors alongside volunteers?
18:08The ones in Hanover don't like the idea.
18:12They think it could demotivate unpaid authors.
18:15But there's no question that challenges are real.
18:18In an age of AI and bots, misinformation and manipulation is rampant.
18:23A growing problem.
18:26Thirty to forty percent of all Internet traffic now comes from automated programs.
18:31Author Torben Friedrich shows examples from ornithology.
18:34Search engine results already include many AI-generated images, often indistinguishable from the real thing.
18:43Even Google can no longer reliably distinguish species.
18:47That makes Wikipedia the most reliable source out there.
18:52Only humans can critically evaluate sources.
18:56I personally believe Wikipedia will survive.
19:02Because we're a large community of people committed to the idea of free, well-documented knowledge.
19:12It's a big group of idealists.
19:17But the platform needs new contributors.
19:20Young authors willing to write and review articles in their spare time.
19:25People who want to help keep Wikipedia relevant and develop it further.
19:29The team in Berlin is working hard to recruit them.
19:34Volunteers who want to ensure that the world's largest collection of knowledge remains a bastion against misinformation.
19:46Are you looking for answers to some of the biggest questions in the universe?
19:50Like, how did galaxies and planets form?
19:54And could there be other worlds capable of supporting life?
19:58NASA is about to begin mapping the universe in greater detail than ever before,
20:02with a brand new space telescope that's been years in the making.
20:06Now it's finally ready, and launch is scheduled for late August 2026.
20:17It's designed to detect light from billions of galaxies.
20:22Discover thousands of new planets orbiting other stars.
20:27And to help clear up mysteries that shapes the evolution and makeup of space and time.
20:34And we really need to do that because we're trying to answer fundamental questions about the universe.
20:39Roman will have a field of view that's 100 times bigger.
20:41So he'll be able to survey the sky 100 times faster.
20:49Instead of, like, focusing on an individual tree, you can see the big forest.
20:58The Nancy Grace Roman Space Telescope, or Roman for short, is named after NASA's first chief of astronomy.
21:06She's often called the mother of Hubble for her tireless efforts to gain approval and funding for the iconic instrument
21:13launched back in 1990.
21:15She was really the person that came up with the plan and the idea to do astrophysics from space.
21:23We wouldn't really have Hubble or Chandra or Compton Gamma Ray Observatory or any of these major space missions without
21:32her vision and leadership.
21:37The space telescope that bears her name carries two powerful instruments behind its 2.4-meter mirror.
21:45The first, called the Wide Field Instrument, is an infrared camera that will allow Roman to survey a field over
21:53100 times larger than Hubble can manage.
21:56And offering a broader cosmic perspective than even the James Webb Space Telescope.
22:02The interesting thing is that Hubble and Webb are, like, looking through a keyhole.
22:09You can get a lot of detail, but Roman will do things differently.
22:12It's like opening the door, and it's going to allow us to map the universe.
22:18The second instrument on board, called the coronagraph, is a complex device for screening out a star's overwhelming glare.
22:26Together, the two tools will provide fresh data on a range of questions.
22:32One of the most tantalizing among them being...
22:34Other worlds, like ours, somewhere else in the universe.
22:39How common is our system like our own?
22:43To date, we have found over 6,000 planets circling other stars.
22:48Roman will dramatically expand our exoplanet catalogue.
22:52NASA expects it to detect 100,000 more in the coming years.
22:57Most of those discovered so far have been very large, as they're simply easier to spot across vast distances using
23:04indirect methods.
23:06But Roman should be able to see some of them directly.
23:10It will take really fuzzy pictures, we hope, of planets around nearby stars with a technique called direct imaging.
23:16So not just using indirect indicators, but actually detecting exoplanets as faint points, even light years away.
23:24We'll block the light from the stars to see the light from the planet either emitted or reflected from the
23:31planet's surface.
23:31And with that, we think we'll be able to see Jupiter-like objects.
23:36What about planets smaller than that, more similar to Earth, worlds that might be like our own?
23:43Detecting those is the primary goal for the European Space Agency mission PLATO, which is scheduled to launch in early
23:502027.
23:52The ESA Space Telescope has 26 cameras, and for years at a time, it will train them on nearby stars
23:59similar to our Sun.
24:01Because it will specifically focus on detecting Earth-like exoplanets up to the habitable zones around those stars,
24:09ESA's PLATO will provide a great complement to NASA's Roman.
24:14Roman will also help researchers see far beyond nearby exoplanets.
24:20The data it harvests should help provide brand new insights into a major cosmic mystery.
24:28Roman is going to be answering big questions about the dark universe.
24:32It has a gravitational pull. We can see its effect on other matter in the universe, but we don't know
24:37what it is, so we call it dark matter.
24:39Dark energy is something else that has the opposite effect. It is driving the universe to expand.
24:44We don't know what it is, but we see it acts like an inflating energy, so we call it dark
24:48energy.
24:51Little is known about the nature of dark matter and dark energy.
24:55Neither has actually ever been directly measured.
24:58But astronomers believe they've played a decisive role in the development of the universe that we can see.
25:05In models, dark matter makes up a vast, invisible web that shapes how galaxies form, move and evolve,
25:13while dark energy powers the expansion we see around us.
25:17Understanding them better would tell us more about why things are the way they are today and where they're headed.
25:24These are kind of fundamental questions about our universe, but they're really hard to answer.
25:31The Nancy Grace Roman Space Telescope should bring us a step or two closer to answering them.
25:45From biocomputing to imperfect information and interplanetary ambitions, that wraps things up this time around.
25:54Thanks for joining us and hope to see you again soon on Tomorrow Today.
25:59Bye for now.
26:09Bye for now.