Biocomputer 'brain in a jar' learns to control robots
If you've seen Robocop 2 or read Victor Pelevin's KGBT+, this news will ring a bell, heralding if not the dawn of technological singularity, then the echo of its heavy tread. Chinese scientists have grown brain cells capable of controlling a robotic body.
Living brain cells, wired as an organoid biocomputer on a chip, are learning to control robots thanks to an open-source intelligent interaction system called MetaBOC. The project aims to return human brain cells to artificial bodies.
What is a biocomputer and biocomputing?
Biocomputing is one of the strangest areas of new technology. It is based on the ability of neurons to perceive the world and provide feedback using electrical signals. Human brain cells, grown in large numbers on silicon chips, can receive electrical signals from a computer, interpret them, and provide feedback. Here is a working example of how this is how the system works.
Biocomputer on the market
This “brain-on-a-chip” system is capable of learning. The concept was first presented by DishBrain project at Monash University in Australia. The researchers grew about 800,000 brain cells on a chip, placed it in a simulated environment, and watched as the cyber-organic mass learned to play Pong in about five minutes. The project received funding from the Australian military and grew into a separate company, Cortical Labs.
In this interview with Cortical Labs Chief Scientist Brett Kagan the idea is clearly stated: even at an early stage, biocomputers based on human neurons learn much faster, consuming several times less energy than modern machine learning chips with artificial intelligence, while demonstrating “intuition, insight and creativity.” After all, our brain, also a kind of biocomputer, consumes only a tiny 20 watts to operate, which is negligible compared to trying to deploy consciousness on the server.
Bottleneck in attempt to build a biocomputer
We've done tests of traditional reinforcement learning. In terms of how quickly a neural network has to look at a number of samples before it starts to learn effectively, it's like night and day. Biological systems, even as simple and clunky as we have now, outperform the best deep learning algorithms humans have ever built. It's pretty wild.
Brett Kagan, Chief Scientist, Cortical Labs
One of the drawbacks, besides the ethical issues, is maintaining an optimal environment for the biocomputer to operate. This means that the grown brain needs to be supplied with nutrient mixtures, liquids, temperature control, and protection from microbes and viruses. The record for the lifespan of the Cortical organoid in 2023 was about 12 months. Well, transhumanism theory gradually develops into practice.
Modern Biocomputer. Merging Human Brain and Cyborg Body
Similar projects are unfolding all over the world. Research at Indiana University, where brain cells were allowed to self-organize into a spherical organoid «Brainoware», to which electrodes were then connected. Or the Swiss startup FinalSpark, which uses dopamine as a reward mechanism for biocomputer chips Neuroplatform.
Now Chinese researchers say they have taken this area of technology to a new level.
Biorobot
The MetaBOC project (BOC stands for brain-on-a-chip) brings together researchers from the Tianjin University Haihe Brain-Computer Interaction and Human-Computer Integration Laboratory with other teams from the Southern University of Science and Technology.
MetaBOC is an open-source software. It is an interface between brain-on-a-chip biocomputers and other electronic devices. The main goal is to give the organoid the ability to perceive the world through electronic signals, interact with the environment using any controls, and learn to solve specific problems. It is quite possible to connect to such a system wireless prosthetics.
The Tianjin team says it uses spherical organoids, like the Brainoware team in Indiana, because the three-dimensional physical structure allows for complex neural connections to form, similar to those in our brains.
These organoids are grown under the influence of low-intensity focused ultrasound stimulation which the researchers say gives them a more intelligent foundation to build on.
The MetaBOC system also attempts to combine different types of intelligence, using artificial intelligence algorithms in software to interact with the biological intelligence of brain cells. It all comes down to the fact that the brain we are used to model of intelligence will be revised almost constantly.
From neural networks to biotechnology
The Tianjin team specifically mentions robotics as a goal for integrating the technology they've created, and provides some rather crude but illustrative images. The biocomputer on a chip, the team says, can now learn to operate a robot by figuring out the controls and attempting to do things like avoid obstacles, track targets, or learn to use its arms and hands to grasp objects.
Since the brain organoid can only “see” the world through electrical signals fed to it, it could theoretically learn to pilot a mini-gundam in a fully simulated environment, allowing it to avoid most crashes and collisions without putting the organic intelligent engine at risk. Although, it is already in critical superpositionlike the brain of any living creature.
The organoids themselves probably won't be that big in the early stages of research. However, they will need all sorts of support equipment, including fluid and nutrient supply lines, pathogen-proof seals, temperature control systems, and shock protection.
Biocomputer and development prospects
In any case, if you build a small robot with the appropriate sensory and motor capabilities, there is no reason why human brain cells cannot learn to control it.
This is a phenomenal time for science and technology, with systems being built on the one hand human brain computeraimed at connecting high-speed computer interfaces and implants in the brain, not without problemswhile projects like MetaBOC are growing human brain cells and integrating them into computers. And the burgeoning artificial intelligence industry is trying to surpass the best of biological intelligence with consciousness on silicon.
Science and technology are forced to come out into the open philosophy of transhumanismas they confront the limits of our understanding. Do brains grown in petri dishes have consciousness? Do AIs have consciousness? Maybe their work is only described dataism? Thus, in the near future, such types of intelligence will become indistinguishable from intelligent beings, giving rise to anarcho transhumanism. What will the ethics be when this happens? Are the ethical foundations different for carbon and silicon intelligence?
Ethics of a Brave New World
This section is an excerpt from a speech by Brett Kagan, Chief Scientist at Cortical Labs
Let's say that these systems do develop consciousness – I think it's very unlikely, but let's say it does. Then you have to decide whether it's ethically right to test them or not. But we already test on conscious beings. We test on animals that I think have some level of consciousness, and we don't have any concerns about that… We eat animals, many of us do it without any concerns, and that's justified.
Frankly, I can hardly believe what I am writing: that humanity is beginning to take the physical building blocks of its mind and use them to create a cybernetic mind capable of intelligently controlling machines.
But this is life in 2024, as we hurtle full speed toward a mysterious technological singularity, the point where AI intelligence will surpass our own and begin developing new technologies faster than humans can.
We are already looming at a point where technological progress – which is already proceeding at an unprecedented rate – accelerates exponentially and becomes vertical, and we lose control of it entirely. In this time frame, we are lucky enough to live in bodies and not be a bunch of cells plugged into a chip in a dish. At least as far as we know.
More materials on the topic of consciousness, intelligence, brain function and the improvement of humans as a species – read in telegram channel. Subscribe to stay up to date with new articles!