An AI Story
The human mind is a great storyteller. This comes from the basic modus operandi of evolutionary symbolic intelligence in humans, as a symbolic narrative creator. In other words, the way in which the mind makes sense of anything, deep in its evolutionary structure, is essentially a narrative. This narrative capacity - the power of symbolic intelligence - evolved in the first place as a new way of we humans making sense of our conscious experience of the world and our self in it.
The narrative is basically a story in which our experiences and the perceivable things of our world must hold together in a consistent narrative about them, as those experiences and perceivable things change in time.
This is a consequence of evolution. It is a consequence of hundreds of thousands of years of the evolution of symbolic intelligence in humans.
Even science is such a narrative, with the peculiar exception that it is a narrative disciplined by scientific method.
So we tell ourselves stories, to make sense of our experience of the world and our experience of self within it.
AI is quite a frontier change in the world. More than that, it's a frontier change in the world that encroaches on the very narratives that we already tell ourselves, about ourselves. And about the nature of the world in which we live. And the reason it does that, is because, literally, it is trained in the first place on the narrative output from humans.
So it's like a giant mirror of our own narrative. But one that is able to play around with the narrative in a new way. And enhance it. As long as it still seems to hold together as a suitable narrative, according to our existing narratives. Otherwise, it's called an AI "hallucination".
Of course, large language models don't hallucinate, according to the original sense of the word. The use of this word tacitly or unconsciously supports a narrative supplied by humans, about AI. Until this happened, the word "hallucination" meant a false sense perception. And of course, a large language model has no sense perception. But hijacking the word "hallucination" quietly conflates what AI is doing there, with something the human mind can sometimes do, making it look as though AI is a kind of mind. And certainly, if we want to be undisciplined in the way we relate to AI, then we may well think of it as a kind of mind.
And that - the ongoing, repeated, subtle suggestion that AI is a mind, at the same time as propagating the story that we know it's not - really, is central to the current human narrative on AI, that is now playing out in the press, and in media comment and programs on the subject.
At no point will anyone from the industry claim explicitly and unambiguously "It is a mind" not least because to talk about "mind" rather than "neural networks" and "intelligence" and "consciousness" and so forth, is not sufficiently scientific or conversant with how modern science views the brain. But nonetheless, much as we might like to avoid it, the word "mind" is what this is all about. The problem is, that humans have what psychologists call "theory of mind". And they are using that all the time in narratives about AI.
This isn't the same thing as anthropomorphism. Commentators are often careful to point out that we shouldn't anthropomorphise AI. But you don't have to be anthropomorphising AI, to be projecting "theory of mind" onto it. In these times, anthropomorphism and "theory of mind" are not quite the same thing.
We know how we built AI, even though we don't know how it works at its core. And yet at the same time, as soon as debate opens up on AI, the subtle, suggestive narrative that it is a mind, creeps in. In no time at all. And the fact that we don't know how it works only plays into this. It leaves the door open for "theory of mind" to try to justify itself in application to AI.
Now some of the latest narratives (October 2026) are pushing very hard to try to establish ambiguity over whether or not a large language model consciously experiences. With reports that we can now probe into its unconscious working, and that it seems to feel pain, and so forth. But of course, all of this is narrative, and "theory of mind" in action. And the narrative is not only "theory of mind". It goes much further.
The more subtle but more powerful narrative in the popular media about "looking inside" the parts of AI that we do not understand, here, is that "what we are talking about is science", when in fact, the popular narrative is pseudoscience. And then that, itself, is in general always quickly conflated with the popular narrative that "we don't know" whether or not there is anything that constitutes consciousness there, because we are still "waiting for the science". This is little short of mentalism.
One of the most amusing parts of this narrative theatre is about science itself. The narrative that science will answer it, eventually. And that we are just waiting for the data, because science and data will answer everything, because science is the alpha and omega of all human knowledge. Including knowledge of the mind. Applying this narrative to the question of consciousness in AI, is completely and utterly anachronistic. It attempts to pass off unnoticed the unexamined idea that at some point in the not too distant future science is going to be able to answer this question. It is a modern form of mythopoeic prophesy, concerning the future of science, dressed up in a disguise of rationality.
When it comes to the question of what consciousness and mind actually are, we also invariably run into the religion-like faith that if we can propose or do an experiment, even when the result could mean anything, it will work towards the scientific knowledge we imagine is going to be forthcoming, of what consciousness or conscious mind is, as a scientific fact. When the truth, all along, is that conscious phenomenology and scientific fact are two completely different things that cannot be collapsed. Two completely different registers of human experience and understanding. We can certainly show some correlations, but any attempt to conflate them results in pseudoscience and/or delusion.
But then, the symbolic mind is a storyteller. It was born that way, out of its evolution. Its fundamental modus operandi is narrative. It is all a consequence of the evolution of symbolic intelligence itself. Of which a large language model, the amplifying mirror, is now the artificial version.
The narrative is basically a story in which our experiences and the perceivable things of our world must hold together in a consistent narrative about them, as those experiences and perceivable things change in time.
This is a consequence of evolution. It is a consequence of hundreds of thousands of years of the evolution of symbolic intelligence in humans.
Even science is such a narrative, with the peculiar exception that it is a narrative disciplined by scientific method.
So we tell ourselves stories, to make sense of our experience of the world and our experience of self within it.
AI is quite a frontier change in the world. More than that, it's a frontier change in the world that encroaches on the very narratives that we already tell ourselves, about ourselves. And about the nature of the world in which we live. And the reason it does that, is because, literally, it is trained in the first place on the narrative output from humans.
So it's like a giant mirror of our own narrative. But one that is able to play around with the narrative in a new way. And enhance it. As long as it still seems to hold together as a suitable narrative, according to our existing narratives. Otherwise, it's called an AI "hallucination".
Of course, large language models don't hallucinate, according to the original sense of the word. The use of this word tacitly or unconsciously supports a narrative supplied by humans, about AI. Until this happened, the word "hallucination" meant a false sense perception. And of course, a large language model has no sense perception. But hijacking the word "hallucination" quietly conflates what AI is doing there, with something the human mind can sometimes do, making it look as though AI is a kind of mind. And certainly, if we want to be undisciplined in the way we relate to AI, then we may well think of it as a kind of mind.
And that - the ongoing, repeated, subtle suggestion that AI is a mind, at the same time as propagating the story that we know it's not - really, is central to the current human narrative on AI, that is now playing out in the press, and in media comment and programs on the subject.
At no point will anyone from the industry claim explicitly and unambiguously "It is a mind" not least because to talk about "mind" rather than "neural networks" and "intelligence" and "consciousness" and so forth, is not sufficiently scientific or conversant with how modern science views the brain. But nonetheless, much as we might like to avoid it, the word "mind" is what this is all about. The problem is, that humans have what psychologists call "theory of mind". And they are using that all the time in narratives about AI.
This isn't the same thing as anthropomorphism. Commentators are often careful to point out that we shouldn't anthropomorphise AI. But you don't have to be anthropomorphising AI, to be projecting "theory of mind" onto it. In these times, anthropomorphism and "theory of mind" are not quite the same thing.
We know how we built AI, even though we don't know how it works at its core. And yet at the same time, as soon as debate opens up on AI, the subtle, suggestive narrative that it is a mind, creeps in. In no time at all. And the fact that we don't know how it works only plays into this. It leaves the door open for "theory of mind" to try to justify itself in application to AI.
Now some of the latest narratives (October 2026) are pushing very hard to try to establish ambiguity over whether or not a large language model consciously experiences. With reports that we can now probe into its unconscious working, and that it seems to feel pain, and so forth. But of course, all of this is narrative, and "theory of mind" in action. And the narrative is not only "theory of mind". It goes much further.
The more subtle but more powerful narrative in the popular media about "looking inside" the parts of AI that we do not understand, here, is that "what we are talking about is science", when in fact, the popular narrative is pseudoscience. And then that, itself, is in general always quickly conflated with the popular narrative that "we don't know" whether or not there is anything that constitutes consciousness there, because we are still "waiting for the science". This is little short of mentalism.
One of the most amusing parts of this narrative theatre is about science itself. The narrative that science will answer it, eventually. And that we are just waiting for the data, because science and data will answer everything, because science is the alpha and omega of all human knowledge. Including knowledge of the mind. Applying this narrative to the question of consciousness in AI, is completely and utterly anachronistic. It attempts to pass off unnoticed the unexamined idea that at some point in the not too distant future science is going to be able to answer this question. It is a modern form of mythopoeic prophesy, concerning the future of science, dressed up in a disguise of rationality.
When it comes to the question of what consciousness and mind actually are, we also invariably run into the religion-like faith that if we can propose or do an experiment, even when the result could mean anything, it will work towards the scientific knowledge we imagine is going to be forthcoming, of what consciousness or conscious mind is, as a scientific fact. When the truth, all along, is that conscious phenomenology and scientific fact are two completely different things that cannot be collapsed. Two completely different registers of human experience and understanding. We can certainly show some correlations, but any attempt to conflate them results in pseudoscience and/or delusion.
But then, the symbolic mind is a storyteller. It was born that way, out of its evolution. Its fundamental modus operandi is narrative. It is all a consequence of the evolution of symbolic intelligence itself. Of which a large language model, the amplifying mirror, is now the artificial version.