Lab-grown mini brains can be used to perform simple math tasks. But the question is how far this organoid technique should go.
Clumps of human brain cells grown in a petri dish in the lab, called brain organoids, have been linked to computers and used to solve mathematical equations. The work is a first step towards using living brain tissue as a form of artificial intelligence. This could raise ethical questions in the future, the researchers themselves note.
In an article dat has been posted online for peer reviewa review by fellow scientists Feng Guo, Assistant Professor of Intelligent Systems Engineering at Indiana University Bloomington, and his colleagues say that they have “created living AI hardware that harnesses the computational power of three-dimensional biological neural networks in a brain organoid.” The paper says that “Brainoware,” as the researchers call it, can learn from training data and that experiments show it can be applied in the real world.
READ ALSO
Will there be people on the moon again in 2025?
In two years, the Artemis 3 astronauts will be hopping across the moon, NASA insists. Does the recent success of Artemis 1 prove that target date after all…
Energy and data guzzlers
Computer-controlled AI is getting better at certain tasks. The driving force behind this is ever-larger and more energy-intensive systems that train with ever-larger data sets. For example, the AlphaGo system, which first beat humans in the game of Go, was trained on 160,000 games, more than a human could play in a lifetime.
People have to play a lot less games to get good and our brains also use a lot less power with about 20 watts. That’s why some researchers think biocomputing — using living brain cells as AIs — is the way forward.
Pong
left in 2021 Brett Kagan from Cortical Labs in Australia and his colleagues found that they had slices of human and mouse brain cells trained to play the computer game Pong. The “DishBrains,” as the team called them, learned much faster than conventional AIs.
Guo’s team didn’t use slices, but three-dimensional structures: brain organoids. When stem cells are cultured under the right conditions, they spontaneously form brain organoids, which do not grow larger than a few millimeters because they do not have blood vessels.
The researchers used these human brain organoids to solve a mathematical equation called a Hénon image, which is difficult to predict because of its chaotic behavior. According to the article, Brainoware outperformed conventional AIs without so-called long-short-term memory (long short term memoryLSTM), but was less accurate than AIs with an LSTM.
Exciting idea
It’s an interesting approach, but it doesn’t show practical applications, says physicist Martin Lellep from the University of Edinburgh in United Kingdom conventional AI to predict Hénon images. “The forecast is not incredibly impressive.”
And while the scientific pre-publication of the study says Brainoware learned over time, it’s not clear how the feedback was provided, Lellep says.
“While their pre-publication needs more detail, it’s an exciting idea to explore,” says Kagan. “Organoids are an interesting next step in using biological neurons for information processing, something we’ve also been exploring with several collaborators over the past 15 months, and they can show many interesting patterns of activity.”
Ethical border organoids
Brain organoids are very small and disorganized compared to real brains, and their computing abilities are quite rudimentary, says developmental biologist Madeline Lancaster from the MRC Laboratory of Molecular Biology in Cambridge in the UK. “I think it’s a bit far-fetched to compare brain organoids with current AI.”
Researchers are trying to create more advanced organoids, for example creating a kind of circulatory system that allows them to grow larger. ‘We certainly want to avoid crossing an ethical boundary. The scientific and ethical community will come together to determine where that limit is before we reach it,” says Lancaster.
“Right now we’re probably years away from real ethical boundaries, but we want to have the discussion now and not just when there’s a conflict of interest,” says toxicologist Thomas Hartung from Johns Hopkins University in the United States, whose team provides Kagan with brain organoids that should have better long-term memory.