There aren’t many computer chips that you have to build a life support system for.
But when you’re combining actual living brain cells with inorganic silicon chips, you can’t feed them just electricity. You actually need to supply everything they would normally get in a fully biological body.
As Hon Weng Chong, the CEO of Australia’s Cortical Labs explains, it’s all about creating computer systems that learn — and that learn faster with less training data. That requires a different approach than standard Intel, Nvidia, or AMD chips, he says.
“What we’ve actually built is a hybrid chip that is comprised of a CMOS sensor, so it’s a silicon chip with a very fine mesh of electrodes. They’re about 17 microns in pitch and there are about 22,000 of them,” Chong told me on The AI Show recently. “And what we’ve done is we’ve taken live neurons that we’ve extracted from mice embryos or we’ve differentiated them from stem cells and grown neural networks on the actual chip surface.”
Why brain cells?
The only machine we know that actually has true intelligence is the brain, Chong says. And the blocks of carbon and protein in brain cells form together and are able to produce computation, he adds. Neurons on silicon chips form synapses by “hybridizing” with the silicon surface. And that means you now have a programmable biological machine. “Because these are electrodes, we can see the electrical activity and also apply a stimulus, a bit of a voltage, and in a sense, now we have a read and write interface into a biological substrate,” Chong says. The first thing these biological machines are doing? Playing ping pong.
Cortical Labs chief technology officer Andy Kitchen says they’re in early stages of teaching their hybrid chips to play ping pong using a stimulus-response methodology. In other words, training these brain-chip machines is more like teaching a child than programming a computer. “Just like any other learning happens through kind of a series of stimulus-response cycles, like learning to ride a bike, we create a very specialized stimulus response cycle in order to induce a specific behavior that we care about,” Kitchen says. Ultimately, the goal is to imitate the complexity that very simple biological machines (animals, creatures) exhibit with — shall we say — astonishingly tiny CPUs.
For example, C. Elegans, a tiny worm or nematode about 1 millimeter long, has a grand total of just over 300 neurons. And yet with that very limited mental capacity, it is able to exhibit interesting behavior: finding food, avoiding danger, reproducing, and carrying on its existence. Flies don’t have large brains either, but they maneuver well, are hard to catch, and have complex behavior. Cortical Labs hope is that biologically-enhanced AI systems would be able to learn complex actions as well: manufacturing, driving, building, cleaning, and so on. And that the biological chips will learn faster. And learn more like we do.
“What we think is going to be really amazing, and we’ve seen this with some of our chips as well, is the fact that these neurons rewire, they actually reprogrammed themselves in order to solve the particular task,” Chong says. “So, I mean, it’s the same thing for us as humans, right? Or even a dog, you teach your dog how to play fetch. You’re changing its environment, you’re changing the stimulus and it’s adapting to it. It’s not reprogramming, it’s not going in there and like rewiring.”
And that’s not done by individually programming every step, as traditional technology requires. Rather, you program a high-level goal or end state, and the system wires itself together to achieve that. Of course, humans have a lot more neurons than nematodes or flies: 86 billion, give or take a few million. Cortical Labs’ current Brain Chip One has between ten of thousands to hundreds of thousands of neurons. In the company’s roadmap, that scales to millions. At which point, assuming the company can both program them and keep them alive, interesting behavior and capability should arise.
“We would see hundreds of thousands of neurons and millions of neurons is certainly more powerful, have more latent power, than an equivalent silicon system,” Kitchen says. One question I had: how do you actually connect biological mouse neurons with silicon computer chips? They don’t, after all, come with a miniaturized version of USB or HDMI. Apparently — brace yourself for the high-level science here — you smear them pretty much like peanut butter.
“The system we use … are these micro electrode arrays and they’re really grids of microscopic electrodes,” Kitchen told me. “And what you do is you basically, I mean, think of it in a simple as possible sense, like smearing peanut butter on a piece of toast, right? You take these neurons and neural progenitor cells and you smear them on top of this electrode grid and there are certain binding chemicals as well, which can cause them to stick better. And these neurons are so close to these electrodes, just physically close to these electrodes, that when they fire you can pick it up.”
It’s not exactly Frankenstein and lightning. But perhaps it’s comforting that something so high-tech could also be so simple. One question remains to be answered: what are the ethical considerations of putting biology and silicon together? And at what level of sophistication do we “create” something that deserves more rights than, say, a washing machine? At the moment, those remain questions without answers. But as this technology is in development, and if it actually comes to market in five years or so, they’ll be worth exploring.
One more note:
If we actually do end up using biological components in our computers, “my computer died” is going to end up having a whole new meaning.