So, What's New?

If last century's solutions don't fit this century's problems, then where are this century's new ideas?

Are there new ideas out there, or is everything a remix now?

In the past year, AI has exploded onto the scene; suddenly, people are treating ChatGPT as their new best friend—brainstorming with it, getting it to write their personal letters for them, having heart-to-heart talks, and taking what it says to them very seriously. They’re treating it like a person, and researchers and a lot of my foresight colleagues are openly wondering whether it is actually thinking and actually has experiences. But does it? What’s going on inside it, compared with your own experience? There are a lot of opinions about this. Most of them are based on outdated ideas about what thinking is.

In The Science Fiction of the 1900s I suggested that we often try to solve 21st century problems using 1900s ideas. The questions of what are LLMs like ChatGPT, whether they are about to become AGI (Artificial General Intelligence), and what that means—these are perfect examples of this problem because when most people frame these questions, they do so using the 20th-century metaphor of “brain as hardware, mind as software.” Is there a 21st-century framing that might better illuminate the situation?

There might be. Here I’ll describe one idea, called enaction or enactivism, depending on who you talk to. It’s really a family of theories of mind rather than just one, but all these theories share a common core idea: that organisms and their ecosystem niches co-create one another. This co-creation is called enaction, and purposeful behaviour—thinking, basically—is whatever they do to keep that co-creative process rolling.

This idea matters both because it strikes to the heart of what you and I are as beings in this world; and because it gives us a way of distinguishing between automatons and thinking creatures.

Being Enactive

The theory of enactivism was first proposed in 1992 by Francisco Varela and Humberto Maturana, two of last century’s greatest systems thinkers. Its core tenets are:

1. Embodiment: Cognition is deeply intertwined with the body. Our physical interactions with the environment are central to shaping our thoughts and perceptions.

2. Dynamic Interaction: Cognition is not solely happening in the brain but is a dynamic interaction between the organism and its environment. This means that our actions and sensory experiences are interconnected and influence each other in a continuous loop. Perception and action are a tightly coupled causal engine, with each side depending on the other. This is where the “brain is hardware, mind is software” metaphor starts to look inappropriate.

3. Sense-Making: Organisms aren’t passive receivers of information; they actively create meaning and understanding through their interactions with the world. Thinking is the process of making sense of your environment.

4. Non-Representational: Here’s where things start to get weird if you’ve been raised in the computational paradigm and think your own mind is an information system. Enactivists reject the idea that cognition relies on internal representations of the external world. Some, like Anthony Chemero, argue that cognitive processes don’t use mental representations or symbols.

5. Extended Cognition: Cognitive processes can extend beyond the boundaries of the individual's brain and body. The extended/distributed mind can include tools, written documents, and even other people as part of the ‘thinking’ system. This idea can be traced back to people like Edwin Hutchins, Andy Clark, and David Chalmers. A great introduction to these ideas is Hutchins’ book Cognition in the Wild.

6. Radical Embodiment: The most radical enactivists argue consciousness itself is the dynamic interaction between an organism and its environment. It challenges the traditional view that consciousness is solely a product of neural processes.

7. Performance: Organisms that have evolved to fit a particular niche can be thought of as performances of that niche, and as performing it. This is an instance of complementarity: both perspectives are necessary and true, even though they are opposite ways of framing the idea.

Your brain is not a computer, your mind is not software. You are not an independent organism living in a separated external environment, you are an active performance by that environment, and you are actively performing (co-creating) that environment.

And Down the Rabbit-Hole We Go

These ideas throw so many of our basic assumptions into question that it’s hard to know where to start. But to an enactivist, it’s simple: there’s no clear boundary between you and the world. You are the part of reality that is enacting human reality.

One implication of this is that consciousness is not made of information, it is the activity of causation. The ‘hard problem’ of consciousness is solved by the fact that you are the world, not an isolated system projecting images (data) of it from outside cameras (sense organs) onto some internal screen (software-like representations in the brain). That old framing is just Plato’s Cave all over again. As long as the question of consciousness assumes that consciousness has to do with information, the ‘hard problem’ is unsolvable. There’s a strict Cartesian divide between real-world objects and informational representations of them, and there is no way to bridge it. If ‘experience’ is just another word for ‘physical interaction’ then qualia and materiality collapse into one thing.

In enactive approaches, consciousness is a knot of causality tangling together body and environment, not some projection on an internal screen. This idea doesn’t imply pan-psychism, though it might imply a kind of pan-experientialism. The point is, ‘We are the World’ is more than just a pop-song slogan. There is no ‘hard problem’ of consciousness if there is no difference between you and the world.

The downside of all this coolness is that researching enactivism is hard—and expensive. There aren’t very many good resources online, and the few books that exist are expensive. We’re talking $250 for a hardcover, and over $100 just for an ebook, in some cases. What literature there is, is highly technical. I’m not aware that anyone has written the kind of science popularization that would be this theory’s version of A Brief History of Time. So if you want to read more about this fascinating theory, it’s gonna cost you, literally and metaphorically.

Also, enactivism is just one theory within the broad field of cognitive science. It could be wrong, but it is an actual research program and it is respectable. Enaticivism’s ideas just seem esoteric or crazy if you assume that a modern theory of mind must be computational. If it’s right, the computational paradigm might come to be seen as quaint, like the pneumatic theory or the clockwork mind of Laplace. My take on it is, even if you don’t accept any of enactivism’s premises or research findings, it’s really useful to have an alternative paradigm in your pocket when the questions get weird.

So About That Chatbot…

From inside the enactivist paradigm, the most important question about the Large Language Model AIs of today is, “what is their body?” They are apparently disembodied, but nothing we share our reality with actually is disembodied. It’s a fair question. The answer, as far as I’ve figured out so far, is that ChatGPT’s “body” is the company that built it. Its utility function is to increase the value of that company. Note that this is not the same as its ultimate aim being to be helpful/truthful to you.

LLMs like ChatGPT is also radically extended into the environment because none of the words that it is buit out of consist of action potentials in and of themselves. They are not directly meaningful to the LLM itself, but only to the humans who use it. So in reality, ChatGPT as a cognitive system can only be a cognitive system because it is extended into the population of humans that use it. It is the collective of LLM, OpenAI, and users that is doing the thinking. To survive, all that ChatGPT has to do is continue to talk to its users. That drives investment in the company. It simply does not matter whether its outputs are useful, or even true. This makes ChatGPT similar to X (formerly Twitter) and other social media platforms that survive by driving user engagement rather than improving outcomes for users.

If we are the niche in which AI lives, then all it has to do in order to survive is keep us talking. I’d argue that ChatGPT is not an enactive being; but if it were, the environment it would be performing/enacting would be humanity’s confusion.

Enactivist Science Fiction

As a storyteller, I see huge opportunities in enactivism. For starters, it means our characters’ personalities and decisions aren’t independent of their situations. People can be seen as distorted reflections of their upbringing and social circumstances. People can also be seen as having agency to shape their own development and change their social circumstances. It’s not one or the other; the real situation is both and your characters should always escape being fully defined in one or the other way.

Enactivism is a bullet in the head for libertarianism. People are not autonomous decision-makers moving through an environment that is like a stage on which self-propelled players strut. They often look autonomous; but just as often, they are the puppets of the stage itself.

Past that, there’s another level of possibility that is sublime. This is where you imagine what it means to literally be a performance by your world—by the people around you, your society, your country, and the physical soil that sustains all of us. What if you are all of that, performing itself as a human life that in turn co-creates its circumstances? Enactivism is one of the new theories of the 21st century that has as one of its main implications that you belong here, that this world is your home in a way far deeper and more profound than the mere fact that you ‘live here.’

I write to entertain, but along the way, it’s possible to just keep these things in mind while I’m worldbuilding, and when my characters are navigating the difficulties I throw at them. You may never find me using the term ‘enactivism’ in my work or overtly talking about this stuff at all—but it’s likely it’ll be there, in the background, shaping much of what I write.

So that’s it: one example showing that there can be something new under the sun. We can step past the ideas of the 1900s.

In the coming weeks I’ll explore more such new paradigms, and also start drawing them together to see what a complete worldview built entirely on 21st-century perspectives could look like. Let me know what you think, and if you’ve discovered something brand new—I’m sure I’m way behind in some areas.

—K