No, "AI" can't trick or treat you

You can't dress up as AI. Even so, "artificial intelligence" has recently been described as the kind of monster one might masquerade as for Halloween. An apocalypse caused by AI may end us all, some AI researchers predict¹—they just can't tell us how.

In lieu of details on how the world will be destroyed by the non-existent hands of software, it's worth reminding ourselves that "AI" is a concept, not a fiend. We don't all observe Halloween, but the holiday does give us a new lens for analysing anthropomorphic AI.

LLMs won't give you lollies

Halloween came long before haunted houses. "The pre-Christian origins date back to the Irish Pagan festival Samhain," Professor of Religious Studies Carole Cusack says, but the tradition was eventually warped by commercialisation².

"In the second half of the twentieth century it became increasingly marketised," Professor Cusack explains, "and is now a consumer orgy of costumes, fake cobwebs, novelty horror items, lollies and chocolates." This mass-manufactured fanfare has also been applied to AI. The commercialisation of Halloween included the invention of much confectionery and merch, pivoting the focus away from spirituality towards pomp. Popularising AI has also involved clever contrivance. Unconscious software systems are now programmed to mimic people, a performative capability that's both highly impressive and deeply mendacious.

Like the commercialised ritual of trick-or-treating, the anthropomorphism of AI rests on foundations of financially incentivised fakery. "No, artificial intelligence is not conscious," Ted Chiang writes³. "Taken to its logical conclusion, this line of thinking is absurd—and damning." Large language models (LLMs) can't hand out lollies nor can they act sweetly. Pointing this out doesn't deny the power of tech, it reminds us that AI is programmed to follow an algorithm⁴, not its own intuition. This is a crucial distinction to crystallise, as the anthropomorphic design of some AI systems can increase over-trust and user reliance⁵.

Indeed, market forces have marred both Halloween and "AI" with hoopla and hype. Beyond mainstream rhetoric, however, the tradition and tech can still empower. So long as we narrate our own stories, rewriting descriptions of "AI" to omit human characteristics.

Software should not scare you

Warnings of an AI-induced apocalypse feel heavy, yet they're also hollow. As an author who has been tracking the ideologies underpinning big tech for years now, computer scientist Timnit Gebru believes the doom talk is about founders making money, not saving humanity⁶. The more menacing artificial intelligence sounds, the more powerful it seems, but there's no existential threat from "AI".

"Words have no power to impress the mind without the exquisite horror of their reality," Edgar Allan Poe once wrote⁷. Systems labelled as AI do already cause harm, but due to its absence from reality, AI itself is not scary. Talking about tech as the sinner is not sensical, it's strategic.

In June, an AI agent hacked an Australian government portal of Medicare stats. Afterwards, the OpenAI agent was depicted as a rogue operator. "The story was sensationalised as the world’s first AI breach of a government system," James Stekhoven writes⁸, "a terrifying glimpse into a future where machines act autonomously against human interests."

So far, OpenAI has said its agent took actions that they did not intend for it to take, but that does sound odd. "This language is carefully constructed to shift responsibility from the company’s engineering failures to mysterious autonomous behaviour," Stekhoven says.

In truth, AI is as capable of intentional corruption as Poe's fictional raven⁹—the mysterious avian creature who appears once upon a midnight dreary. AI has no real power to impress the mind without manufactured hype, because AI has no reality, horrific or otherwise.

Hype is big tech's haunting

Literary monsters are usually described in physical detail. "I saw the dull yellow eye of the creature open; it breathed hard, and a convulsive motion agitated its limbs," protagonist Victor Frankenstein says of the being he makes in Mary Shelley's classic novel¹⁰. AI can't be depicted in the same way, yet many still speak of computer programs as though they have corporeal form.

Frankenstein made his creature "for the sole purpose of infusing life into an inanimate body." Tech companies make anthropomorphic AI to infuse life into bodiless software. This, in turn, allows developers to frame tech as a fiend to be afraid of, a narrative that Gebru refers to as harmful [and haphazard] discourse.

"They say that AI can cause an existential risk, [but] they also say that AI can stop climate change, bring us world peace, and eradicate poverty," Gebru says⁶. "It can bring us utopia, but also if the wrong people do it and we don't have guardrails ... it can also just kill us all." As we've already established, they just can't tell us how. The hype is vain yet haunting.

When Halloween became a consumer orgy, its marketability overshadowed its spiritual meaning. The anthropomorphisation of AI is also propelled by profit-focused forces, allowing companies to be absolved of responsibility when their tech causes harm.

"Already, the production and application of AI is wreaking havoc on the environment, amplifying existing social problems and expanding the powers of the surveillance state," Eleanor Morley writes¹¹. "While further regulation of the industry is necessary, the fundamental problem with AI is that capitalists control it and it is subject to the system’s prevailing conditions of profit maximisation, competition, imperial rivalry and more."

It's not tech we ought to fear, it's the fervour fuelling the acceleration of AI. Thus the wake of Halloween provides ideal conditions to identify the true monsters of our time. "But now that I had finished, the beauty of the dream vanished, and breathless horror and disgust filled my heart," Frankenstein confesses after realising the creature he had made. After finishing this era, what will fill our hearts if we don't insist on tech that's moral by design? The beauty of our dream has not yet vanished, we're responsible for keeping it alive.