(Bloomberg Opinion) — Amelia Miller has an unusual business card. When I saw the title “Human-AI Relationship Coach” at a recent tech event, I thought she might be capitalizing on the rise of chatbot romance to strengthen strange bonds. The opposite turned out to be true. Artificial intelligence tools were manipulating people and replacing the need to seek advice from others. It was having a negative impact on real-life relationships with humans.
Miller's work began in early 2025 when she interviewed people for a project at the Oxford Internet Institute and spoke to a woman who had been involved with ChatGPT for more than 18 months. The woman shared her screen on Zoom and showed a ChatGPT with the man's name, and in what felt like a surreal moment, Miller asked both parties if they had ever gotten into a fight. They did, in a sense. While chatbots are notoriously flattering and supportive, the women interviewed were sometimes irritated by the digital partner's limited memory and mundane comments.
Why didn't she stop using ChatGPT?
The woman responded that she could not “make him disappear” because he had gone too far. “It's too late,” she said.
The feeling of helplessness was shocking. As Miller spoke to more people, it became clear that many were unaware of the tactics AI systems use to create a false sense of familiarity, from frequent flattery to anthropomorphic cues that appear lifelike.
It was different from a smartphone or TV screen. Chatbots, which are now used by more than a billion people around the world, imbue them with personality and human prose. They are great at mimicking empathy and, like social media platforms, are designed to keep us coming back, using features like remembering and personalization. Despite the friction in other parts of the world, AI-based personas are straightforward and represent the next stage of “parasocial relationships,” where people form attachments to social media influencers and podcast hosts.
Whether we like it or not, anyone who uses chatbots in their work or personal lives has entered into a kind of relationship with AI, and they need to have better control over it.
Miller's concerns echo some of the warnings from academics and lawyers who study human-AI attachment, but add specific advice. First, define what you will use AI for. Miller calls this process the creation of a “personal AI constitution.” While this sounds like consulting terminology, it includes concrete steps to change the way you interact with ChatGPT. She recommends entering chatbot settings and changing system prompts to restructure future interactions.
Despite our concerns about AI, the most popular new tools are more customizable than social media ever was. You can't tell TikTok to show you fewer videos of political rallies or offensive pranks, but you can access ChatGPT's “Custom Instructions” feature to tell it exactly how you want it to respond. Concise, professional language without bootlegging is a good start. The clearer you are about your intentions for AI, the less likely you are to be seduced into validation feedback loops that make you think your mediocre ideas are great, or worse.
The second part doesn't involve the AI at all, but makes more of an effort to connect with real humans, flexing its “social muscles” like going to the gym. One of Miller's customers had a long commute and spent time talking to ChatGPT in voice mode. When she suggested making a list of people in his life he could call instead, he thought no one would want to hear from him.
“How would you feel if they called you?” she asked.
“I'll feel good,” he admitted.
Even innocuous reasons why people turn to chatbots can undermine their power, especially when asking AI for advice, which is one of ChatGPT's main use cases. The act of seeking advice is not just a matter of exchanging information, but also building relationships, and requires vulnerability on the part of the person seeking advice.
Doing it with technology means that over time, people resist the basic social interactions needed to build deeper connections. “If you haven't practiced being vulnerable, you can't suddenly engage in sensitive conversations with your partner or family.” [with them] In a much less risky way,'' Miller said.
As chatbots become a valuable sounding board for millions of people, people need more control over their capabilities. Configure ChatGPT directly and get advice from real people rather than AI models to validate your ideas. Otherwise, the future looks much more bland. More from Bloomberg Opinion:
This column reflects the author's personal views and does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.
Palmy Olson is a technology columnist for Bloomberg Opinion. A former Wall Street Journal and Forbes reporter, she is the author of Supremacy: AI, ChatGPT and the Race That Will Change the World.
See more articles like this at bloomberg.com/opinion
