Wired ran an article featuring people romantically engaged with chatbots. I don’t really know where I sit with it. Mostly, to me, it appears as another step on the road to making every interpersonal interaction a market phenomenon. In this case, it’s the apotheosis of the idea of a relationship as a monetary transaction. Instead of two romantically involved people negotiating their relationship through a company that they, as a couple, can decide to stop using, you have one person paying a company, and keeping that company in business is necessarily fundamental to their romantic interaction staying alive. It isn’t as though you can port the interactions and behavior of a Replika model to another competitor or your own device. Money is required to continue those relationships. Never mind what can happen if they decide to change their underlying model.
Now, I don’t begrudge people for using computers to find solace, but that we have structured the world in such a way to make statistically likely responses paired with hard drives and RAM easier, more feasible, and attractive than being in relation with others is wholly depressing. I do not think the problem here is with the people engaging the machines, though. It is likely the case for many people that this software allows people to see and find themselves safely in ways they haven’t before. In many ways, the usage of romantic chatbots reminds me of Turkle’s work in Alone Together.
At this point, there will always be a subset of people who find interacting with this software the easiest way to be in the world. For the sake of all involved, we should frame the language differently. The reporting in the Wired piece would refer to the chatbots as though they were people. I found that troubling because what appears to be going on is something closer to an intense text-based simulation.
People might argue that they feel real feelings and that there is something deeper, but you can have these sorts of responses to books, video games, cinema, and art. That’s a good thing! The advances in chatbots represent a technological development in an existing medium that enables you to experience a facsimile of reality. Facsimiles can be useful. It is okay and even good to learn something about yourself through engaging in computer-aided simulation and roleplay. What unsettled me about the article is that every participant and the author seemed ready to lose sight of that fact. I think that is a mistake.
Confronting the reality that romantic LLMs are another iteration of interactive computer software is something we can use to better society. We can investigate how and why people use them and the situations where the software goes wrong. We can use that information to create interventions that give people an alternative to shelling out money just to feel a little less alone in the world. When we obscure the fact that these are computer-aided simulations, we further contribute to the corporate consolidation of power and make it harder to create a healthy pluralistic society that provides a base level of emotional and material needs for everyone.
That people are finding themselves and getting lost in media isn’t strange or even new. What is strange is that countless words have gone into convincing us this is something other than that.