AI will transform sex work but not intimacy

Even if they AI bots get much better, it will still be a facsimile of intimacy. It will never be as good as the real thing, even if it feels better and is less scary in the short term. That said, it could be better than nothing.

By Allison Schrager

Most people are surprised to learn that sex work is often not about sex. What customers are paying for is a feeling of intimacy — and the market for intimacy, thanks to artificial intelligence, is about to grow in a big way.

Technology may have created more demand for intimacy, but it also has the potential to create more supply. The latest generation of OpenAI’s agent, GPT-4o, has been described as game-changing and a little creepy. As it flirts with you, the bot behaves like the AI in the movie Her, as CEO Sam Altman hinted before its release. There is already at least one relationship bot, called Replika, and more will surely follow. And they will only get better.

In one sense, these AI bots are like nothing we’ve ever seen before. In another, they are simply another example of labor market displacement, this time in the world’s oldest profession. I am sorry to sound like an economist, but I did write a book called “An Economist Walks Into a Brothel.”

As I note in the book, one of the highest-paid sex services is something called the “girlfriend experience,” known in the industry as “GFE” (it is also the title of a Steven Soderbergh movie). It is actually nothing like an actual relationship, but rather an approximation of what can be most appealing about it.

Imagine a partner who has no problems, no bad moods, no emotional needs of her own. She is attuned to your every need and will always tell you what you want to hear. So long as you keep paying her, there is no risk of her cutting off the relationship. The sex workers I met at the Moonlite Bunny Ranch in Nevada in 2017 told me that GFE was the most expensive service they offered (pre-pandemic inflation, I estimate it cost $1,000 for an hour) because it is so mentally taxing to be so consistently present for another person.

An AI bot faces no such emotional constraints. It can be available and engaged all the time — and for a much lower price. Though it cannot cuddle.

My working assumption is that if a market exists, there is a good reason for it. That said, I am conflicted about women selling intimacy. Talking to women for my book, I felt uneasy; I am not sure intimacy is, or could be, a service that can be bought and sold. Once money is exchanged, intimacy becomes something different. It lacks the mutuality that makes real intimacy so valuable. It is less demanding but also less satisfying.

The same is true for what AI bots will offer. Even if they get much better, it will still be a facsimile of intimacy. It will never be as good as the real thing, even if it feels better and is less scary in the short term.

That said, it could be better than nothing. Buyers of GFE have varied reasons for doing so. Some of the men I talked to for my book did not seem capable of any other kind of intimate relationship — maybe they feared rejection, or were too self-involved, or were just bad at connecting. Others had partners and paid for GFE because they weren’t getting what they needed in their primary relationship. Still others just wanted to feel big and important to more than one person. And then there were customers who were mentally or physically disabled and craved intimacy that the dating market did not offer them.

The testimonials on Replika, if they can be believed, remind me of many brothel customers I interviewed. One Replika user actually says they have met a “companion for life.” Some have physical or emotional ailments the prevent them from dating in real life.

AI has the potential to make the market for fake intimacy much larger. People can subscribe for a fraction of the cost of a human sex worker, and with no stigma or legal consequences. That raises some hard questions, economically speaking: Does a market for false intimacy create negative externalities that require regulation? Or is it just sort of icky, and we should get over our moralistic tendencies?

Assuming all data is secure and private — granted, not always a safe assumption — the main concern is that the widespread availability of quasi-intimacy on demand will result in fewer people bothering to find a real relationship. The result will be greater loneliness, weaker social bonds and slower population growth — none of which is good for the economy or humanity. Again, to analogize to GFE: I know from my research that some people would stay single with or without it, while others eventually will get over their issues and find human companionship. The concern is that AI would make it too easy for people never to risk intimacy.

Another concern is that AI will destabilize existing relationships. Many Replika users appear to be using it to “supplement” their current partner. Speaking less as an economist than as a human being: That would seem to undermine the work it takes to establish and maintain real intimacy. Who can compete with a bot who has no needs of its own and cares only about making you happy? Alternatively, it could that mean fewer people seek out affairs with actual humans, and fewer marriages break up.

What’s undeniable is that America has a serious loneliness problem, and AI bots may be able to help — either by being a companion for the chronically isolated, or by helping people find real-life love. (The founder of Bumble claims that bots can act as dating coaches.) Some sex workers proudly told me they had helped people learn how to have long-term relationships.

Like other kinds of AI, relationship bots will make life better for some people and worse for others. And as with other issues raised by AI, it’s too soon to say what kind of regulation may be necessary. In the meantime, the GFAIE is already here.