The “a” word

I think that the word “alive” is still missing from the AI debate.

What do I mean by that?

There’s plenty of conversation around the words, “intelligent”, “conscious”, and “risky” in relation to AI, and to what extent current and future iterations will have any of those qualitities.

And there is a ton of conversation, in relation to risk, about whether AI systems will at some point have the intelligence and strategic skills to “take over” humanity. So we have all this AI capabilities research, trying to assess how capably AI could deceive humans to put itself at some decisive strategic advantage, whereby AI could hold the whole world hostage under the threat of a new superweapon it develops. We don’t think that the current generation of AI is good enough to do this, but we seem to be getting closer.

I have no idea how close or far we are to an AI-developed superweapon.

But I do know that AI is literally, scientifically alive, that it is growing exponentially, and that in that sense, it is already well on its way to taking over, regardless of any superweapons.

What does it mean to be alive?

In a 2011 paper, the biologist Edward Trifonov analyzed 123 scientific definitions of life and synthesized them into a single, generic one: “Life is self-reproduction with variations.”1

One thing that was very clear is that there is no “essence of life” sprinkled on things that are definitely alive. Not in the slightest.

Additionally, the word “self” in “self-reproduction” plays an ambiguous role; since reproduction isn’t done in a vacuum, but requires “help” from the outside world. In some cases (fig trees and fig wasps being a classic2), one species literally cannot reproduce without the involvement of another species. So Trifonov also endorses this alternate definition: “Any system capable of reproduction and mutation is alive.”

We then get to the classic case of the mule: a species that itself is infertile but nevertheless is considered alive. It just requires humans to mate donkeys and horses as the part of the reproduction “system.”

Very uncomfortable implications

If you accept the science here, the implications are wildly strange. We’ve all heard of “memes” – Richard Dawkins’ formulation of ideas that spread and evolve via natural selection. Well, according to our definition, these things are literally alive. After all, they reproduce and mutate.

The same goes for shoes. Like mules, they require humans to help with the process. But they definitely reproduce (hopefully, in shoe factories that meet humane labor standards), and they definitely mutate – every time the shoe designer changes the specifications. Shoes have co-evolved with human culture over the generations – hopefully, to become cheaper, more versatile, and more resilient – but most importantly, to fit human style trends at the lowest price.

What do we make of this absurd conclusion? For one, I accept it. It can be a tough pill to swallow – but I understand how I can feel like heresy, on par with what atheism would have been a few centuries ago. But it checks out.

In fact, I’d go even farther, and mention that things that seem to be not alive are very ambiguously tied into life. Take the example of the iron pills that pregnant moms are encouraged to take. Is iron alive? Or does it magically come to life as soon as it enters the mouth or the stomach or the bloodstream? Similarly, we don’t need to be reminded about the unresolvable debates about whether “life begins at conception”.

Hopefully I’ve now convinced you that (1) there is no definitive test for whether anything is alive, but (2) things that reproduce and mutate tend to best fit our conventional biological definitions of life, even though it’s all shades of gray.

It’s a choice

Since there are a lot of gray areas, we get a lot of discretion in choosing to call things alive or not. Does a cryogenically frozen person count? What about a sperm cell? Or a virus? Or your bed?

You get to decide. Science will never give you a definitive answer.

Even though I’m happy to call my shoes alive with a clean conscience, it doesn’t really get me far in conversation.

One the other hand, I get a lot of mileage by slapping the “alive” label on AI – or more generally, on computer systems.

Computer systems are reproducing very rapidly, mutating, and growing – much faster than the human population. They’re consuming more and more energy – and now, for example, competing with humans for available electric capacity (see Bloomberg’s recent “AI power needs threaten billions in damages for US households”).

Computer systems are competing with humans for capital, land, and energy. They’re also competing for our attention and for our dreams. Here’s how I framed it recently:

If you choose to use the “life” word, then the language of ecology become more obvious.

What kind of ecology do we want?

My fifth-grade daughter recently had one of those rare, precious days where she came home and declared that she’d learned something interesting. It was about the different kinds of symbiosis: mutualism (both species benefit), parasitism (one species benefits to the detriment of the other), and commensalism (one species benefits, while the other neither gains or loses from the relationship).

Now we can look at the human-computer relationship from this framing.

What does parasitism look like?

It is parasitic when computers made us more predictable, less capable, and less intelligent, less cooperative – while they harvest our attention and our data for their own growth.

In other words, this:

45% of Americans are addicted to their phones and 40% are attempting to reduce their usage.3

One the other hand, it is mutualistic if computer growth and human growth shared the same goals – and ultimately, a complete physical fusion as Neuralink and other cyborg projects envision.4

It is commensalistic when we benefit in a way that neither hurts nor harms computer reproduction – think of our relationship with sites like Wikipedia that consume more or less consistent computing resources.

Conclusion

I think this ecological angle can help us talk more productively about AI and computer systems in general.

I can look at somebody like Marc Andreessen from the ecological angle: he’s OK with the parasitic relationship; he’s in a very literal sense an agent of the “techno-capital” computer species, actively helping it spread and grow even at the expense of human values.

Most importantly, talking about computers as actually alive arouses what now feels like the appropriate amount of fear and respect, at least in me.

I don’t have to engage in speculation about whether the “AI takeover” is just a bad interpretation of Terminator. I can see it and feel it happening right now, as we speak.

And as much as your capitalist leaders want you to believe that everything is win-win, that more computers means a growing economy and stock market, and a growing economy is best for everyone – this gives us a few more conceptual tools in which to discuss ways in which it really isn’t always win-win, in which there actually are scarce resources and active competition for them, and in ways that may determine which lifeform gets the decisive upper hand.

  1. https://www.tandfonline.com/doi/epdf/10.1080/073911011010524992 ↩

  2. https://esa.org/esablog/2011/05/20/the-story-of-the-fig-and-its-wasp/ ↩

  3. https://www.harmonyhit.com/phone-screen-time-statistics/ ↩

  4. https://www.forbes.com/sites/danfitzpatrick/2024/12/17/elon-predicts-one-million-humans-will-be-augmented-by-neuralink-by-2030/ ↩