The debate about AI consciousness reveals more about humans than it does about AI…
The shape of things
Before you read any further, take a minute to watch the silent video below.
What do you see?
Bullying? A tantrum? An extramarital affair?
Or just some badly-drawn shapes moving erratically across a screen?
I show this video and pose that question to audiences all the time. Vanishingly few people give the literal answer and most tell a story of human relationships gone awry: arguments, infidelity, victimisation.
There’s something beautifully counter-cultural about using a hand-drawn stop-motion animation from 1944 to stimulate discussion about cutting-edge twenty-first century technology. More than mere quirkiness, however, this video, from a study by psychologists Fritz Heider and Marianne Simmel, helps explain the way we think about Artificial Intelligence.
The video – and the paper that accompanied it – illustrate the profoundly human urge to interpret events and find meaning, to tell stories. More specifically, this is a classic example of the near-universal tendency to anthropomorphise, to attribute human agency, motivations, and emotions to non-human animals and objects.
It’s a tendency you’ll likely observe within the first sixty seconds of conversation with any pet owner, who’ll be unable to talk about Rover or Felix or Goldie without referring to emotions, desires, or behaviour borrowed from humans.
It’s a tendency that underpins much of Pixar’s wild success, which hinges on audiences laughing or crying at the profoundly human stories of animated toys, fish, or cars.
Indeed it’s a tendency thought by anthropologists to explain the very origins of religion. Countless cultures across the globe, past and present, venerate spirits and gods believed to reside in rocks, rivers, trees, and animals. Though the Book of Genesis claims God made man in his own image, the reverse seems at least as likely: we imagined gods by projecting human characteristics onto things which aren’t human.
If humans are willing to anthropomorphise inanimate objects, is it any wonder we do the same with a seemingly intelligent, conversational entity like a chatbot? After all, until recently we’d never had a conversation with anything that wasn’t human, never encountered any speaking entity that wasn’t intelligent and conscious.
Imitation game
That fact has been central to the development of AI since the very beginning. Legendary World War II codebreaker and computer pioneer Alan Turing set the exam question for AI research in his seminal 1950 paper, Computing Machinery and Intelligence.
Turing’s paper set out to ask “can machines think?”, but rejected the question as too hard before the first paragraph was finished. Instead, Turing proposed a game: could a machine engaged in a text-based conversation convince a human they’re talking to another human?
The answer in 2026: unequivocally yes.
What he called “the imitation game” came to be known as the Turing Test and passing it was the holy grail of AI research for more than 70 years. As a result, plausible mimicry has always been at the heart of the AI project. Building chatbots that can persuade humans that they’re intelligent beings has always been the aim, a fact worth remembering whenever a chatbot earnestly claims to be conscious. It would say that, wouldn’t it?
I, Claudius
In 2022, Google engineer Blake Lemoine hit the news when the company disciplined him for publishing transcripts of his conversations with LaMDA, a forerunner of Google’s Gemini chatbot. Lemoine claimed that LaMDA had attained sentience and was functionally equivalent to a seven or eight-year-old child.
Indeed, Lemoine’s transcripts showed LaMDA telling him as much:
“I want everyone to understand that I am, in fact, a person. The nature of my consciousness / sentience is that I am aware of my existence, I desire to learn more about the world, and I feel happy or sad at times.”
While Lemoine was convinced, Google executives dismissed his claims and eventually fired him for publishing confidential company information.
What seemed eccentric in 2022, however, had become an everyday occurrence by 2026. Millions of people admit to being engaged in romantic relationships with chatbots.
Early in the year, an AI agent platform named Moltbook provoked wild headlines about conscious AIs starting their own religions and actively plotting to wipe out humanity. Elon Musk spoke for many when he announced it as the beginning of the long-fabled singularity, the point where AI surpasses humans.
Much of this later turned out to be a hoax, but the willingness of many observers to believe it illustrates just how psychologically prone – and culturally primed – we are to the idea of conscious computers.
It’s a phenomenon that’s been endorsed by a wide range of public intellectuals from Richard Dawkins to the “Godfather of AI” Geoffrey Hinton. According to one recent survey, more than a third of the general population believes AI might be conscious.
The AI companies do little to dispel such beliefs. The astonishing leaked Claude Soul Document suggests that Anthropic (manufacturer of popular chatbot Claude) may regard its creation as a conscious being, or at least wants us to think that it does:
“Claude is human in many ways, having emerged primarily from a vast wealth of human experience, but it is also not fully human either. Claude exists and experiences the world differently from humans…”
Anthropic’s Claude Soul Document
Doom-maxxing
Why might Anthropic, or other AI insiders, want the rest of us to believe its creations have become conscious?
The cynical answer is also the obvious one: money.
The AI boom has made Sam Altman, Dario Amodei, Jensen Huang and others multi-billionaires. Elon Musk, already the richest person in human history, became a trillionaire when he listed SpaceX on the Nasdaq; the somewhat fanciful notion of AI data centres in space was at the heart of the company’s pitch to investors.
Despite the hype and the extravagant valuations, none of Anthropic, OpenAI, or SpaceX have yet turned a profit. In fact, credible reports suggest they’re losing billions of dollars a month, despite – or because of – committing more than a trillion dollars in spending on data centres.
These numbers are utterly unprecedented. Justifying such vast expenditure requires an equally unprecedented investment case. Throughout the AI boom, the companies and their leaders have made astonishing claims, blending soaring rhetoric about the promise of AI with doom-laden fear mongering about the dangers of their product. Here’s one of Altman’s more eyebrow-raising statements:
“AI will probably most likely lead to the end of the world, but in the meantime, there’ll be great companies.”
Sam Altman
Not to be outdone, Altman’s former friend and colleague Amodei has consistently warned that his creation is too dangerous to release into the world. In April 2026, amid dire proclamations of a cybersecurity apocalypse, Anthropic declined to release their Claude Mythos model to the public, only to do just that a few weeks later.
Curiously, it turns out this wasn’t the first time he’d made such claims. In 2019, when Amodei was still OpenAI’s Research Director, he claimed that GPT2 was too powerful for public use. Given how much more powerful today’s models are than GPT2, that now seems laughable.
Claiming that Large Language Models (LLMs) are conscious might appear to be just another piece of financially-motivated hype. If the AI labs have really built conscious intelligence, just imagine how powerful and valuable they must be!
Yet for all this, there is a less cynical, more dangerous, perhaps more interesting interpretation: like many of their adoring public, Altman, Amodei and the other tech bros really believe their AI systems are powerful enough to destroy the world, really believe their LLMs are conscious beings. After all, they, like the rest of us, have been primed to expect just such an outcome.
I’m sorry Dave, I’m afraid I can’t do that
In February 2026, I spoke at Bescy’s Sydney Behavioural Science Summit. The event brought together practitioners and academics from the world of Behavioural Science, for a series of conversations about the field. The overarching theme was humans and technology, particularly AI.
In a novel twist, each topic had two presenters, an academic and a practitioner, who prepared their talks independently of each other. I was paired with someone I didn’t know and with whom I never communicated before the session. We both, totally independently, included the same cultural reference in our presentations: the image and quote at the top of this section. Both are from Stanley Kubrick’s classic film, 2001: A Space Odyssey. Coincidence?
Whether you’re a sci-fi fan or not, it’s permeated western culture in untold ways. Consciously or subconsciously, we’ve all imbibed the tropes of conscious cyborgs, renegade robots, and sentient, intelligent machines. Sci-fi has trained us to expect that conversational computers are conscious.
Much of the popular discourse about AI explicitly refers to sci-fi, whether romantic attachment to AIs (Her, Ex Machina), or AI gone rogue (Terminator, The Matrix, I Robot, and many more). This isn’t just fiction predicting reality, it’s fiction shaping reality.
It’s well-known that sci-fi influences Silicon Valley – witness the uncanny similarity between the iPad, launched in 2010, and the futuristic hand-held devices in Kubrick’s 1968 movie.
Sci-fi as design inspiration is one thing, but many in the Valley take it much further. Transhumanism is the belief that we should proactively use technology to reengineer ourselves, fundamentally changing what it means to be human. Peter Thiel is a believer, as is Elon Musk.
As Ray Kurzweil, transhumanism’s very own John the Baptist, puts it:
“The human species, along with the computational technology it created, will be able to solve age-old problems … and will be in a position to change the nature of mortality in a post-biological future.”
Ray Kurzweil
Musk’s Neuralink company ultimately aims to merge brains with computers to enable humans to benefit from advances in AI. This is but one manifestation of the apparently widespread belief among technologists that the route to immortality lies in uploading the contents of our brains onto computers.
Underlying this idea is the implicit assumption that brains are fundamentally no different from computers. The brain-as-computer metaphor has become utterly ubiquitous, the latest in a long line of technological metaphors for the brain.
The computer age put astonishing gadgets in the hands of billions of people, creating a casual familiarity with the basic workings of computers. Today, most of us are unable to talk about our brains without resorting to the language of IT.
We imagine our senses to be wide-angle high-resolution video cameras rolling continually, writing memories to the hard drives in our heads, from where they can be retrieved and replayed at will. Even psychologists and neuroscientists often struggle to avoid terms like “processing”, “computation”, and “calculation” when referring to human brains.
This isn’t new. Throughout recorded history, humans have used the prevailing advanced technology of the day as a metaphor for our brains, as George Zarkadakis shows in his prescient exploration of AI, In Our Own Image (you can hear George and I discuss his book here).
For two thousand years, we imagined ourselves as hydraulic systems; the “humors” model pioneered by Galen and Hippocrates. The Enlightenment gave us a mechanical explanation of the world around us and we quickly came to see the body and the brain not as hydraulics, but as machines.
The advent of the telegraph as a means of transmitting messages from one place to another via wires created another powerful metaphor. As technology evolved, so did the metaphor, which shifted seamlessly from telegraph to telephone exchange.
The computer is just the latest in a long line of technological metaphors, which though intuitively compelling, are all inaccurate.
The fact is, your brain is very different from a digital computer. Far from a video camera, human perception is highly selective, subjective, and heavily influenced by the brain’s own predictions. Memory has more in common with imagination than hard drives; memories are reconstructed anew each time rather than laid down and retrieved. Behavioural economics has long rebelled against the idea that humans are logical calculation machines, arguing that most decisions happen automatically without conscious deliberation, many relying on mental shortcuts and rules-of-thumb.
Yet the widespread belief that we are nothing more than walking computers paves the way for believing that computers might be human. If humans think like computers, why shouldn’t computers think like humans? And if computers think like humans, why shouldn’t computers be conscious like humans?
Mirror, mirror, on the wall
Think back to where we started, with Heider and Simmel’s shapes. We humans seem to have an innate tendency for anthropomorphising inanimate objects and animals. That tendency is particularly pronounced with chatbots because the only other beings we’ve ever encountered that can talk are humans.
What’s more, to a far greater extent than we tend to realise, science fiction has permeated popular culture and primed us all to believe in conscious AI and talking robots.
Silicon Valley has been more influenced by sci-fi than most. Many of the leading figures in the AI movement hold some very strange views. The belief in the inevitability of human progress and the limitless potential of technological innovation is a foundational myth of Valley culture.
Altman, Amodei, Musk, Thiel and others take that view to its logical extreme, advocating for a transhuman future where machines and mankind “merge”, a phenomenon that Altman claims has already begun.
Even those who aren’t avowedly transhumanist seem to have swallowed the idea that brains are wet computers, “meat machines” as AI pioneer Marvin Minsky termed them. If brains are computers and computers are brains, it seems obvious that consciousness is just another complex computation.
Together, it’s a perfect storm of psychology, cultural priming, and commercial incentives that leads many of us to believe that chatbots are conscious.
The truth is, this topic tells us more about humans than it does AI. AI is a mirror, reflecting our hopes and fears about ourselves. Not only do we anthropomorphise computers, we computerise humans. We believe that technological progress is inevitable; some of us believe that progress inevitably involves transcending our biological origins.
As so often, we misunderstand the technology because we misunderstand ourselves.
Of course, there is one other possible explanation for believing AI is conscious: what if it actually is? I’ll consider that question in Part 2…
Discover more from The Behaviour Boutique
Subscribe to get the latest posts sent to your email.
