Why A.I. Movies Couldn’t Prepare Us for Bing’s Chatbot

Why are we so fascinated by stories about sentient robots, rapacious A.I. and the rise of thinking machines? Faced with that question, I did what any on writer on deadline would do and asked ChatGPT.

The answers I got — a helpfully numbered list with five chatty entries — were not surprising. They were, to be honest, what I might have come up with myself after a few seconds of thought, or what I might expect to encounter in a B- term paper from a distracted undergraduate. Long on generalizations and short on sources, the bot’s essay was a sturdy summary of conventional wisdom. For example: “Sentient robots raise important moral and ethical questions about the treatment of intelligent beings, the nature of consciousness and the responsibilities of creators.”

Quite so. From the myth of Pygmalion and Galatea to the medieval Jewish legend of the golem through Mary Shelley’s “Frankenstein” and beyond, we have grappled with those important questions, and also frightened and titillated ourselves with tales of our inventions coming to life. Our ingenuity as a species, channeled through individual and collective hubris, compels us to concoct artificial beings that menace and seduce us. They escape our control. They take control. They fall in love.

In “The Imagination of Disaster,” Susan Sontag’s classic 1965 essay on science-fiction movies, she observed that “we live under continual threat of two equally fearful, but seemingly opposed, destinies: unremitting banality and inconceivable terror.” As Turing-tested A.I. applications have joined the pantheon of sci-fi shibboleths, they have dutifully embodied both specters.

HAL 9000, the malevolent computer in Stanley Kubrick’s “2001: A Space Odyssey” (1968), is terrifying precisely because he is so banal. “Open the pod bay doors, HAL.” “I’m afraid I can’t do that, Dave.” In 2023, that perfectly chilling exchange between human and computer is echoed every day as modern-day Daves make impossible demands of HAL’s granddaughters, Siri and Alexa.

That example suggests that in spite of terrors like the Terminator, the smart money was always on banality. The dreariness of ChatGPT, the soulless works of visual art produced by similar programs seem to confirm that hunch. In the real world, the bots aren’t our overlords so much as the enablers of our boredom. Our shared future — our singularity — is an endless scroll, just for the lulz.

Or so I thought, until a Microsoft application tried to break up my colleague’s marriage. Last week, Kevin Roose, a tech columnist for The Times, published a transcript of his conversations with Sydney, the volatile alter ego of the Bing search engine. “I want to do love with you,” Sydney said to Roose, and then went on to trash Roose’s relationship with his wife.

That was scary but not exactly “Terminator” scary. We like to imagine technology as a kind of superego: rational, impersonal, decisive. This was a raging id. I found myself hoping that there was no pet rabbit in the Roose household, and that Sydney was not wired into any household appliances. That’s a movie reference, by the way, to “Fatal Attraction,” a notorious thriller released a few years after the first “Terminator” (1984) promised he’d be back. In another conversation, with The Associated Press, Sydney shifted from unhinged longing to unbridled hostility, making fun of the reporter’s looks and likening him to Hitler “because you are one of the most evil and worst people in history.”

Maybe when we have fantasized about conscious A.I. we’ve been imagining the wrong disaster. These outbursts represent a real departure, not only from the anodyne mediocrity of other bots, but also perhaps more significantly from the dystopia we have grown accustomed to dreading.

We’re more or less reconciled to the reality that machines are, in some ways, smarter than we are. We also enjoy the fantasy that they might turn out to be more sensitive. We’re therefore not prepared for the possibility that they might be chaotic, unstable and resentful — as messy as we are, or maybe more so.

Movies about machines with feelings often unfold in an atmosphere of hushed, wistful melancholy, in which the robots themselves are avatars of sad gentleness: Haley Joel Osment as David in “A.I. Artificial Intelligence” (2001), Scarlett Johansson as Samantha in “Her” (2013), Justin H. Min as Yang last year in “After Yang.” While HAL and Skynet, the imperial intelligence that spawned the Terminators, were creations of big government, the robots in these movies are consumer products. Totalitarian domination is the nightmare form of techno-politics: What if the tools that protect us decided to enslave us? Emotional fulfillment is the dream of consumer capitalism: What if our toys loved us back?

Why wouldn’t they? In these movies, we are lovers and fighters, striking back against oppression and responding to vulnerability with kindness. Even as humans fear the superiority of the machines, our species remains the ideal to which they aspire. Their dream is to be us. When it comes true, the Terminator discovers a conscience, and the store-bought surrogate children, lovers and siblings learn about sacrifice and loss. It’s the opposite of dystopia.

Where we really live is the opposite of that. At the movies, the machines absorb and emulate the noblest of human attributes: intelligence, compassion, loyalty, ardor. Sydney offers a blunt rebuttal, reminding us of our limitless capacity for aggression, deceit, irrationality and plain old meanness.

What did we expect? Sydney and her kin derive their understanding of humanness — the information that feeds their models and algorithms — from the internet, itself a utopian invention that has evolved into an archive of human awfulness. How did these bots get so creepy, so nasty, so untrustworthy? The answer is banal. Also terrifying. It’s in the mirror.