Some of you may be familiar with a classic joke from Twitter:

SCI-FI AUTHOR: In my book I invented the Torment Nexus as a cautionary tale

TECH COMPANY: At long last, we have created the Torment Nexus from classic sci-fi novel Don’t Create The Torment Nexus

With that in mind, let’s talk for a moment about AI and the unnerving behavior of its leading apostles. Most of my intellectual betters tell me to stop worrying about AI. “The fears are overblown,” they say. “It’s not going to become Cyberdyne, or cause mass unemployment.” And I will grant that losing my job to a machine is the least of my worries. (Perhaps that is due to how unimpressed I am with AI’s attempts at conservative opinion-writing. If ChatGPT starts cranking out columns slyly referencing old Norm Macdonald “Weekend Update” gags, that’s when I’m pulling a Sarah Connor.)

AI chatbots are not going to replace your plumber, your electrician, your doctor, or even your lawyer, for that matter — though they can genuinely assist the last two in the course of their work. The calculus is rather simple: As long as you can get sued for doing a job incorrectly or negligently, people aren’t going to rely on the say-so of a friendly bot at the other end of an open-ended learning algorithm as their backstop. Perhaps some clerical workers are threatened by AI, and if so then I am sincerely sorry about the remaining student-loan debt from their gender-studies degrees. (As I wrote last year: If we formally agree to start the clerical purge with DEI administrators — who apparently use AI already — then frankly I’m in.)

But nevertheless, most of the people I consider my actual intellectual peers are worried sick about AI, and what primarily alarms them are its sociological and psychological effects. One need only look at the disastrous effect it is having on higher education for evidence of this: Not only are badly formed students who farm their work out to bots turning into young adults who devote their ample free time to activist and identity politics, but eventually, they will usher us into a generational cycle of incompetent and ill-prepared professionals. And the reason the idea of “woke AI” rewriting history is such a political hot button is precisely that we understand most people to be lazy sods who don’t bother with serious research and reach for the quickest brand-name-reliable answer. (These days, it’s invariably Wikipedia for most.)

What worries me most is that AI might one day become yet another — and perhaps the most powerful — technological tool of alienation in a society already atomized nearly to pieces. Those of us raised before the iPhone era have difficulty grasping how the smartphone and smartphone culture have already created an entire generation of emotionally fragile children, with social impulses and incentive structures that seem utterly warped and alien to anyone over the age of 40. Infinite and immediate content, panopticon peer pressure, and an entirely new social-status game have permanently warped the younger generation. Now let’s introduce the predominant/everyday use of AI, with the promise that one day, after we’ve tweaked this algo enough, it can also become your special friend to replace the ones you never tried to meet in real life, you sad lonely shut-in!

And it turns out that I am right to worry about this. Because here comes Sam Altman, leading apostle of AI and the founder of Open AI — the company that kicked off the “AI boom” by bringing ChatGPT to the world — to make America’s skin crawl. A few days ago we learned that Altman tried to get actress Scarlett Johansson to reprise her “character” in the 2013 Spike Jonze film Her (about which more later) as one of the voices — presumably the default one — for ChatGPT’s voice option. “He told me that he felt that by my voicing the system, I could bridge the gap between tech companies and creatives and help consumers to feel comfortable with the seismic shift concerning humans and AI. He said he felt that my voice would be comforting to people.” When Johansson declined, Altman not only went ahead with an exact Johansson soundalike voice instead, but made his intent obnoxiously obvious by tweeting out to the world the single word “Her.”

First of all, I want to salute Johansson for saying no to Sam Altman and Open AI. I am certain that, given how flush with cash they are, they offered her more money for the use of her voice than most actresses will make in an entire career. She said no despite how buzzy and glamorous an emergent technology AI is among movers and shakers. This would have been the easiest paycheck to earn in the world: Sit down in a booth and let someone sample your voice for a few days.

Why did she decline? (Leave aside for a moment just how sleazy it was that Open AI tried to use Johansson’s voice, or its style, even after being rejected — for those suspicious of Silicon Valley business ethics, this was fine vindication.) This brings us to what most unnerves me about Silicon Valley’s proselytization for AI: Its most vociferous apostles seem for all the world to be driven by psychologically bizarre, socially damaged desires. In Johansson’s statement, she says she declined the offer to recreate her character from Her “after much consideration and for personal reasons.” That strikes me as a polite way of saying, “Did you actually watch this film? Did you understand the point of this film?”

For Her is not a love story — it is a cautionary tale. Set in the near future (really just an excuse to dress everyone in awkward ties), it stars Joaquin Phoenix as a lonely and embittered soon-to-be divorcé. As part of his unfulfilling job, he gets a “virtual assistant” with his newest operating-system upgrade: “Samantha,” a learning AI voiced by Johansson that exhibits such naïvely ingenuous lovability as an algorithm new to the world that the alienated and unhappy Phoenix begins to fall in love with it. And, seemingly, “she” with him as well.

It is a marvelous and memorable film — Spike Jonze’s best work by far as both director and writer, with a truly humane, beating heart at its core — so I will not spoil the plot for you from there. But, as those who have seen it know, the film is designed to make you uncomfortable with AI, not more welcoming of it — for AI is incapable of loving you, or even truly befriending you, at least not in the way that people crave and require. Her makes a direct appeal to its viewers to take their heads out of the clouds (and, by analogy, drop the smartphone from their hands) and re-engage with the human reality of the world around them, rather than getting lost in a world of artificial intelligence and pleasures both ultimately illusory and inevitably heartbreaking.

So how the hell could Sam Altman have gotten all of this so wrong? Go back to the “Torment Nexus” joke I opened with. What kind of an utterly dense brick of a person could watch Her and think, “Yeah, I definitely want my AI to remind me of the algorithmic plague that seduced society away from everything except their earbuds before cruelly ghosting planet Earth.” One presumes Open AI has already moved on to option No. 2: getting Hugo Weaving to reprise his role as “Agent Smith” from The Matrix.

So while I am grateful that Scarlett Johansson understands the emotional import of the film far better than Sam Altman does, this episode really does suggest that people like him not only do not understand why people harbor such reservations about AI, they embrace AI for precisely those reasons. Mark Zuckerberg really wants you to live in the Metaverse. Altman clearly seems to fantasize about ChatGPT becoming your late-night AI girlfriend. These are weird people whose psychological priorities are very different from yours and mine, and as long as their aim is the dissolution of modern social bonds, my fears about AI are not about its intellectual capacities (though I remain wary of those). My fears are about its capacity to warp us as humans, and the place that its apostles wish to see it take in our lives.