Recently, scientists at the University of Minnesota created a fully synthetic cell that feeds, grows, and replicates like a biological cell. Dr. Adamala and her team made what they describe as “an incredibly wimpy organism” that must be fed and takes 12 hours to reproduce. I find this fascinating for considerations of artificial consciousness, both metaphorically and by extension. Let’s dig in.
First: is this synthetic cell life? Professor Elani of Imperial College London is cited in the article as saying that it is not “life”, but is a “genuine milestone on the road” towards it. He is quoted as saying, “To my mind, this is a real advance in the long-running effort to ask whether chemistry can be organized so convincingly that we begin to call it life.”
Note the linguistic arc there: chemistry organized in such a way, and presenting such properties (I assume), that we “call it life.” This seems parallel to how I frame the question of phenomenal consciousness in the conclusion of my book. My thesis is that when a digital system presents the functional characteristics of consciousness, we will reach a point where we begin to call it conscious:
Another way of putting it: functional consciousness is engineered, and phenomenal consciousness is emergent — not from continued engineering, but from the embedding of functionally conscious entities within the social world of beings who are already phenomenally conscious. We assign it. Whether that assignment reflects an underlying reality or merely our own projections is, perhaps, a question that cannot be answered, the real chimera of Dennett.
One distinction, though, is that Adamala organized chemistry together through artificial processes; it’s the same physical substrate as bioengineering, the same chemicals and physics. Artificial consciousness as I frame it could be built on a different physical substrate: silicon rather than carbon, circuits rather than veins.
Taking a step back: how we define “life” seems potentially as complex a philosophical question as how we define “consciousness”, and orthogonal to it although inherently related. The article on Adamala’s work simplifies it to something that can feed, grow, and replicate. The concept of “feed and grow” has inherently biological ties to it, though, and
There is a significant wing of consciousness scholarship, one that has received quite a lot of ink and attention lately, that seemingly requires a biological substrate and foundation before the label of consciousness can be attributed. Jaeger et al. articulate what they see as irreconcilable differences in the agency and problem solving abilities of “organismic” entities as compared to algorithmic ones. Seth, who holds a “biological naturalist” view, in “The Mythology of Consciousness” emphasizes the complexity of the brain as something far beyond what AI can replicate or simulate.
I’ve written my response to these thinkers previously; it centers on the notion of abstraction, and the level at which comparisons are made. Any theory of consciousness is looking for certain structural or behavioral characteristics present in what we currently think of as conscious systems; and as soon as something is written down, given enough time and resources, it can be simulated with silicon. As I see it, the important question is utility -- how useful is a theory of consciousness, and for what purpose?
The cynical purpose of asking whether AI is conscious is to garner media attention, and with it the food of technology production engines: investment capital, and developer and partner interest. In my view, the only true purpose in such a determination is whether and to what extent to afford moral rights to an artificial system. That is, ultimately, a societal question, and regardless of the advances of scholarship it will be answered by collective societal understanding and action.
The article on artificial life uses a functional lens: the synthetic cell feeds, grows, and replicates. In my writing, I offer a functional lens on consciousness: a slope rather than a binary, and the more a system presents abstraction and resilience, the more conscious it is. And then I wrap that up in the same way Dr. Elani does, by looking for societal validation of sufficiency: when we as a society see a close enough functional match, we will call it conscious.