Even the smartest AI likely won’t be “alive.” Here’s why.

There is a lot of discussion these days about mimicking consciousness in a computer. From the simulation hypothesis to concern about an impending AI apocalypse, many people talking about one of the core functions of human life (i.e., experience) becoming instantiated in silicon. But below this question lurks a deeper issue. 

Is life itself, sentient or not, something that can be represented in a machine? In other words, is life computable? Whatever life is, can it be reduced to a series of mathematical operations carried out by a computer? Today, I want to introduce you to a remarkable thinker who asked this question more than 30 years ago and whose answer remains important, influential, and very contentious.

Rosen’s claim

Robert Rosen was a theoretical biologist. Trained at the University of Chicago in 1959, he held positions in a variety of institutions. Widely admired for his brilliance and scientific creativity, Rosen pioneered what now might be called a complex adaptive system approach to living systems. For Rosen, living systems were physical, but physics alone could not describe what separated them from the non-living. According to Rosen, the difference was not some special vital essence that exists outside of physical law. Instead, he argued, it was the organization of the organism that demanded a description that rocks, comets, and black holes do not require. Lacking a proper account of organization, Rosen claimed that “the wonderful edifice of physical science, so articulate elsewhere, stands today utterly mute on the fundamental question: What is life?” 

For Rosen, the answer to this question required looking at the definition of cause. He began by taking creative license with Aristotle’s famous distinction between material and efficient cause. Material cause refers to the physical substances or components that make up an object. Efficient cause refers to the factors that bring something into existence and sustain it, such as the processes and interactions that create and maintain a living organism. 

The difference between these causes, in Rosen’s view, can made clear by thinking about a tree and the question “why?” The material cause of the tree, the material answer to why the tree, is all the carbon, oxygen, and other substances from which the tree is built. The efficient cause answers a different kind of “why” question. Rosen saw this kind of cause in terms of constraints, or guiding relations, in the processes that create and maintain the living tree. 

What makes life different?

In terms of physics, there is really nothing other than the natural laws governing the movement of atoms in the tree. Rosen argued this was fine for understanding the material cause but not the efficient cause. That’s because what is unique about living systems is that they are closed to efficient cause. In mathematics, a system is closed if operations in the system lead to more members of that system. For example, when I add two real numbers, like 5 and 2, I get another real number: 7. Since Rosen was trying to build a mathematics of living systems, he was saying that living systems create their own efficient causes. They are self-creating and self-maintaining. That’s what makes life different.

To understand what self-creating and self-maintaining mean, think about the membrane of a single-celled animal. By allowing only certain molecules to enter the cell, the membrane is what allows the organism to metabolize and stay alive. But only by staying alive can the cell create the processes that create and maintain the membrane. The cell creates the processes and products (the membrane) that create the processes and products that create … all in what some people call a “strange loop.”

This closure to efficient cause was central to Rosen’s ideas about life. It was, for him, what set life apart from other physical systems. Most of all, it was what set life apart from machines. Machines always have their efficient cause created by someone else (i.e., humans). Even if you came up with a machine that could fix itself, it would be you who came up with that plan not the machine itself.

Most important for today’s discussion: In a famous 1991 book Life Itself, Rosen provided a mathematical argument based on category theory to show that systems with closure to efficient cause could never have their essential features captured by a computer. The most general version of a computer is Alan Turing’s famous universal machine. It’s composed of long tape made of squares, with each square containing an instruction. There is also a “head” that marches along the tape, reads the instructions, and performs an operation in response (moving forward or back, rewriting the square or not). Since no Turing machine can create its own read-write head, it is clearly, in one sense, not closed to efficient cause. However, Rosen went even further, arguing that systems closed to efficient cause could never be captured by a Turing machine. 

In other words, life is not computable — it’s not a machine.

It’s a pretty amazing claim, but is it correct? Since Rosen published his argument, a small cottage industry of papers has arisen seeking to affirm or disprove the claim. So far, the answer is: It’s not clear. The fact that Rosen’s radical claim has yet to be definitively disproven says something in itself, but only time will tell. I, for one, am rooting for Rosen’s argument. I believe the machine metaphor for life is not only deeply flawed but dangerous in a variety of ways. But what matters even more than the eventual fate of Rosen’s claim is his articulation of life’s closure to efficient cause. 

Like the idea of autopoiesis and other recognitions of life’s essential “strange-loopiness,” self-creation and self-maintenance nail something truly unique about living systems that, one way or another, science must wrestle with.