Dr Melzer, can we have a fulfilling relationship with a robot?

Long gone are the days when computers communicated with us like cold, distant programs. Instead, they are becoming more and more human – and more intimate. OpenAI is reportedly planning to let ChatGPT write erotica in the future, while xAI has created "Ani", a chatbot programmed to be a "companion" in anime style, complete with a short skirt. Several companies are competing with chatbots that imitate certain characters – fictional, historical or contemporary. While even relatively sterile AI voices can already hold convincing conversations today, these companion apps emphasise the programs' apparent "social" capabilities: they ask how the day was, laugh at jokes and provide emotional reassurance.

The technology is still young, even in the context of rapidly developing technology. Yet the media and the internet are already filled with reports of people losing themselves in this illusion of partnership. How do psychologists feel about this rapid development? Dr André Melzer is an assistant professor of psychology at the University of Luxembourg and, in addition to media psychology, focuses on the phenomenon of anthropomorphism, i.e. our tendency to attribute human traits to inanimate objects. This becomes all the easier when these are programs that convincingly "play human", making it increasingly easy for us to see more than just a machine in them.

Lëtzebuerger Journal: There is a lot of media talk about romantic or friendly relationships with AIs. Is that actually a real phenomenon, and how widespread is it really?

Dr André MELZER: We first need to talk about parasocial relationships – these are relationships that resemble real ones, but are not one. The core of this inherently one-sided relationship, be it with a pop star or TV presenter, has always been the illusion of reciprocity, which promotes emotional commitment.

The big and revolutionary difference from the past, however, is that modern AI chatbots offer a genuine feedback loop. They respond to us, and increasingly, they respond with emotionality. The relationship thus shifts from a static, one-sided connection to an adaptive, interactive one, creating an even stronger illusion of mutuality. This interactivity is the foundation on which such relationships could even begin to develop.

Whether they actually do so on a large scale, however, is the subject of ongoing research, and there is still relatively little of it. It's all still a bit too new. What we have seen so far is a strong focus on the target groups of millennials and Generation Z as well as a conspicuous geographical localisation in Asia. Most of the current publications from 2024 and 2025 come from China, Taiwan or Korea, which is probably related to the generally higher technology acceptance there. But even these are just case studies, mostly describing symptoms of dependency related to ChatGPT, but these are isolated cases. A reliable, quantitative body of evidence that allows for solid conclusions about the broader population doesn't yet exist.