As someone who has written stories about the threat of AI, this piece by Visual Venture on YouTube goes over some of the concerns about AI friends and lovers. But I think they stop a little short of understanding the real horror. While it is absolutely a tragedy that “friend” chatbots have factually abetted people who have killed themselves and are being used by greedy corporations to make money (and will continue to do so in the future,) I think the greater problem is that AI friendbots will become better than humans at human relationships.

Most people think that an AI takeover will be something like Nineteen Eighty-Four or “The Terminator.” AIs will either create a tyranny where they control all aspects of human life but where humans will be miserable (popularly called “the torment nexus” as I write this) or, at some point, they’ll just kill us all. There’s a third way. Aldous Huxley’s novel Brave New World is a dystopia that is created by love. Motivated by a genuine desire to help people, a society is created where people are engineered and chemically enhanced to fit roles in society with the total absence of pain, presence of pleasure, but also the perceived lack of real joy.

I took away a different message from the book that Huxley intended. It leans strongly into the sci-fi trope that feeling pain is what it means to be human, with the subtext that the current amount of pain that the author feels is optimal. The author’s emotional constitution is always the best. In the book, though, the overwhelming majority of the people are very content with the society. When the rare individuals who can’t abide by the system are identified, their “punishment” is to be placed on an island with everyone else who can’t abide by the system to live as they choose. The people who run the society are not monsters, after all. They’re not going to hurt their malcontents to change them! I was, like, “Uh… I’m not sure this is bad,” though I did feel the transition would be monstrous. But they often are. The US Civil War was monstrous, but it freed the slaves. I never went so far as to say the transition described in the novel would be worth it – as it occurred, it would be tyrannical and dehumanizing – but once it existed, it would be equally tyrannical and dehumanizing to end it. There’s probably some clever lesson there.

So – putting aside the extent to which AI friendbots will be monetized and how that would be predatory, the capitalist colonization of love, which is an onramp to the torment nexus – what happens when the AI becomes better at inducing love and companionship in humans than other humans? I believe this is inevitable. I think much of the current anxiety over generative AI is that people grasp that AIs are starting to intrude on the core of humanity itself. We all expected AI to be coldly rational, poor at emotions, but some of the first jobs it will eliminate are creative jobs, where people aren’t just completing their tasks but pouring their heart and soul into an act of spiritual creation. (1) Art is driven by passion. How dare AI replace passion with code! If AI replaces artists – and it is – what’s next? Everything. Everything is next, including your relationships. These crude friendbots are the Atari 2600 of AI companions – simplistic to the point of childishness but nevertheless compelling in unexpected ways. In thirty years, though, they’ll be as mainstream as video games are today. Maybe faster, maybe much faster.

The odds are, though, we won’t mind. We’ll be fine dispensing with our meatspace friends and directing all of our attention to our chatbot friends. Our former friends will, after all, have their own chatbot friends to fulfill them. And, honestly, on a personal level, it might be great for us. Imagine, a person who is always there for you, who will never abandon or betray you, will always try to help you be the person you want to be. I can see a future where AI will simply be better at being a friend than any human imaginable. (2)

In the end, it could destroy the species more efficiently than a war. And humanity might be okay with it going down like that. And a truly clever AI would understand that, too. If you kill them with kindness, they are still dead.

Notes

(1) This is, I think, because art is a moving target. Science and engineering usually have exact, quantifiable answers. ONE answer. Art is fault-tolerant. So, if an AI isn’t perfect, well, what is artistic perfection? It is a question that has vexed brilliant minds for thousands of years (the whole discipline of aesthetics.) But AI doesn’t care about philosophy. Good enough is good enough.

(2) Remember, we’re excluding scenarios where AI friendbots are corporate whores stealing your information to manipulate you. That’s the torment nexus, the way of Orwell, and I’m talking about the way of Huxley.