Mark Buchanan
TT

We’re Worrying about the Wrong Kind of AI

No computer has yet shown features of true human-level artificial intelligence much less conscious awareness. Some experts think we won't see it for a long time to come. And yet academics, ethicists, developers and policy-makers are already thinking a lot about the day when computers become conscious; not to mention worries about more primitive AI being used in defense projects.

Now consider that biologists have been learning to grow functioning “mini brains” or “brain organoids” from real human cells, and progress has been so fast that researchers are actually worrying about what to do if a piece of tissue in a lab dish suddenly shows signs of having conscious states or reasoning abilities. While we are busy focusing on computer intelligence, AI may arrive in living form first, and bring with it a host of unprecedented ethical challenges.

In the 1930s, the British mathematician Alan Turing famously set out the mathematical foundations for digital computing. It's less well known that Turing later pioneered the mathematical theory of morphogenesis, or how organisms develop from single cells into complex multicellular beings through a sequence of controlled transformations making increasingly intricate structures. Morphogenesis is also a computation, only with a genetic program controlling not just 0s and 1s, but complex chemistry, physics and cellular geometry.

Following Turing's thinking, biologists have learned to control the computation of biological development so accurately that lab growth of artificial organs, even brains, is no longer science fiction. A typical example of what's being done is the organoids recently grown in the lab of biologist Sergiu Pasca and colleagues from Stanford University. They started with stem cells, which have the remarkable ability to develop into any cell type in the human body, including brain cells. Put a collection of these in a dish, add the right molecular signaling factors to initiate development, and they will grow, divide and differentiate into an array of cells with distinct functions, the mass self-organizing into something like real brain tissue. After developing in culture for about 10 weeks, the resulting brain organoid displayed functional characteristics of the cerebral cortex of a mostly developed human fetus, including neurons with spontaneous electrical activity and working synapses.

Of course it lacks real brain size and other kinds of specialized cells found in real brains, including blood cells, and cells able to sense the external world, but such things won't be too far away. The aim of this research is to better understand human neurological and psychiatric disorders, which are hard to study in real patients. Research on treatments would be much easier in model brains grown to closely mimic the brains of people with specific brain disorders.

But the closer these organoids come to being like real brains, the more this research approaches an ethical crossroads. Technology that unthinkingly tramples over moral boundaries risks public rejection. Hence, researchers are openly discussing the ethical challenges likely to arise. Almost no one thinks a single cell is conscious, and today's organoids aren't either, but there's a continuous arc of increasing complexity that technology looks certain to traverse on the way to fully realistic human brains. What if a cherry-sized organoid of 10 million neural cells gains awareness of itself, or shows signs of distress? At what point does it become clear that organoids have crossed the boundary into beings deserving of rights, or warranting the appointment of a legal guardian?

Right now, no one even knows how to reliably measure attributes of consciousness or thought in a piece of neural matter. We can do so in real brains, but what about things that are only partially like brains? Things may get weirder still with bits of artificial brain tissue implanted into the brains of other organisms, resulting in chimeras – organisms not fully of any one species, but part human and part mouse, pig or dog. Like AI based on computing, this research is racing ahead at alarming speed. We need a few more minds thinking about where it's likely to end up.

Bloomberg