Each week we find a new topic for our readers to learn about in our AI Education column.
Can mankind create a better computer than nature?
Welcome to AI Education, where this week we’re taking a step out into the weirder side of technology. We’ll be exploring biological computing, where living cells, proteins and tissues are used as computers—marking another connection between high-level computing and life itself. Let’s keep in mind that there’s already a biological component to most computing—most computers aren’t completely autonomous, there’s almost always a person interfacing with the machine via a keyboard and mouse or with voice commands.
We should also acknowledge first that artificial intelligence itself was born out of attempts to build machines capable of thinking and behaving like living creatures. Our discussions of machine learning and neural networks explain how much of what we today consider artificial intelligence uses human neurology and human thought as its model. Of course, the artificial intelligence we’ve built thus far does not really mimic human thought. Today’s AI-driven robots are built to physically resemble humans and animals that were products of natural evolution.
But today’s topic, biological computing, approaches the nexus between machine and life in a very different manner. Living cells—particularly neurons and other cells involved in the central nervous system—can be programmed to behave very much like computers. Instead of wires and transistors, biological computers use proteins, DNA, biochemical or bioelectric signals and other means to convey information and perform computations.
What Are Biological Computers
There are different, very specific definitions of biological computers floating around the internet, but we want to use biological computing as an umbrella term that refers to any use of biological cells, tissues or products in order to store or transport information, or to perform computations. We think it sounds like science fiction, but not only are there are already biological computers in use, but scientists have also developed multiple approaches to biological computing.
For example, the story that brought us to this topic broke in March, when Cortical Labs announced a commercially available biological computer, CL1, which combines lab-grown human neurons with silicon hardware not unlike that of a typical computer. We like to think of it as a cyborg on the level of living cells and integrated circuits.
We’re going to talk about several ways in which biology and computer science are intersecting, because it’s not just one story.
Why Create a Biological Computer?
At first, biological computing sounded to us like a mad scientist fantasy, some kind of computer science-meets-Mary Shelley monster in the making, but there are good reasons that technologists and scientists are interested in harnessing the power of the human brain—and life itself—for computing. For one thing, in all the history of computing technology, humanity has only just recently created technology with the reasoning capacity of a human brain. Brains are also capable of a massive amount of parallel—meaning simultaneous—processing, the kind of processing necessary for producing intelligence.
As it turns out, life, and neurons in particular, is incredibly efficient. We’re talking about restarting nuclear power plants to power our AI ambitions with traditional computing, but life can convert its own energy directly from food—or even sunlight and heat. The most powerful supercomputers still require rooms or buildings—entire data centers—to support them, while, over the millennia, life has figured out how to reason pretty well in what is essentially a hunk of meat weighing no more than few pounds on average. We’ve proven to be pretty good at making more life. And the waste? We already flush it down the toilet.
Types of Biological Computers
The CL1 biological computer (which is available for a surprisingly lean $35,000 a pop) is an example of wetware, where living tissue is integrated into a digital system. A silicon computer chip interacts with human neurons within the computer via electrical impulses. Cortical Labs found that through such interactions, lab-grown neurons could be trained to complete tasks and exhibit behaviors—it’s still a far cry from an AI “brain in a jar.”
Most wetware systems, however, use the interaction between a silicon chip and neurons for neurological and pharmaceutical research. That’s actually Cortical Labs stated purpose for CL1, though the hardware could be adapted for other uses. In the future, the technology could help restore function to the injured and disabled, and may come to replace traditional integrated circuit technology. Currently, these systems are not easy to produce and maintain over long periods of time.
Nanobiotechnology, a very different kind of biological computing occurs on the molecular level. Scientists have found ways to store and process information using cells, proteins—even DNA molecules. DNA neural networks have already been developed. Scientists built a biological computer within a single cell of E. Coli that could solve simple mazes. We think of this technology as similar to the tools being used to rewrite genetic code—technology that may, someday, help edit many diseases and conditions out of the human existence.
A third kind of biological computer, related to nanobiotechnology, is implanted into a person or organism to monitor processes or produce a therapeutic effect.
Wait. What Was That About a ‘Brain In a Jar?’
Of course, this is AI education, so let’s talk about what this has to do with AI. The promise of running artificial intelligence using a biological computer is obvious, as this kind of biotechnology could mitigate or eliminate a few of the major obstacles to implementing and advancing artificial intelligence.
And artificial intelligence will certainly advance, as will biological computers, but we can’t help but think about some of the potential implications of the technology.
A wetware computer as we now know it is a programmed clump of human neurons, called a organoid, attached to a silicon chip. Without some instruction and use—some training—these neurons remain, essentially, a disorganized mass—but as they are “used” and taught, they do take form, but it is a form without consciousness. As we start to deploy future iterations of artificial intelligence on biological computers, at some point, those clumps of human neurons are going to attain some level of self-awareness.
