‘I see ChatGPT as a wise companion or advisor’

When ChatGPT was launched in November 2022, Jeronimo got excited. “I was amazed by its capabilities,” he says. A technology enthusiast, he comes from Colombia and obtained his bachelor’s in Economics & Business Economics a few months ago. At first, he used it to explain concepts he didn’t fully grasp in class. 

Later, when OpenAI gave the chatbot the ability to remember things users had said before, the use of ChatGPT spread from his academic life to his personal life. “I used it to plan my finances, manage my portfolio of stocks, look for a house, and find a job. And then I started using it in my relationship. I would ask: ‘Should I have said this or that?’” 

In the end, the relationship didn’t work out “for all sorts of reasons”, and ChatGPT also helped Jeronimo decide when to break up. “It was a form of guidance, not like it made the decision for me. But it certainly helped me see things from a more objective, logical standpoint. Sometimes, we just get driven by emotions,” he explains.

Today, Jeronimo sees ChatGPT as a “wise advisor or counsellor”. He says he doesn’t follow the chatbot blindly, but he finds that “it can give you better advice than most humans if you give it the proper context.” He consults AI for decisions big and small, from potential business ideas to how to maintain a healthy diet and whether to take certain medicines when sick. “It’s just part of my life.”

“It’s like having an objective third party”

Although he wouldn’t describe himself as a “heavy ChatGPT user” as Jeronimo does, Connor, a third-year Media & Culture student, agrees that the chatbot can be a useful decision-making tool. “It helps me make more informed decisions,” says the half-Dutch, half-Scottish student. 

Connor sees AI as an extension of his own decision-making. In his view, this is not a passive act, as if one were consulting an oracle. He argues that taking time to write his thoughts into the chatbot offers a moment of reflection for him. 

He also thinks ChatGPT is “great” for problems with your family, girlfriend, or friends: “It’s like having an objective third party.” 

A few months ago, he got into an argument with some friends and asked ChatGPT if he was right or wrong. He could have asked his mother for advice, “but she’d probably say something like: ‘Oh, forget about those friends!’”. The idea of opening up to other friends about it also felt awkward, so ChatGPT seemed like the best option. “Besides, I could do it from my bed,” he recollects. Connor was surprised when the chatbot took his friends’ side. “It was eye-opening."

The Media & Culture student, Connor Frame. Photo: Ivar Pel

“You may miss opportunities to learn”

Femke van Veelen-Van Haaren, a student psychologist at UU, understands why students are using AI in this manner. “This is a phase in which many experiences happen for the first time: new relationships, new responsibilities, and new choices.” All this novelty feels daunting at times. 

“Chatbots always have an answer, which can feel very reassuring, especially when you feel uncertain,” she says. But there is a caveat: they may also give users the illusion that there is always one clearly correct choice, when life is often messier than that.

To her, the issue isn’t so much whether students use ChatGPT as how they use it. In her view, the platform can be beneficial if used thoughtfully.  “If you mostly ask ‘What should I do?’ instead of ‘What options are there?’, you practice independent judgment less.” Another thing to keep in mind, according to her, is that decisions are not only about logic, but also about values; what matters to you.

She believes that chatbots can be helpful for organising one’s thoughts before talking to someone, for example. However, she would discourage using them to hand over decisions in one’s private life, such as determining what to do or who is right.

The student psychologist stresses the importance of learning to tolerate uncertainty: “If you get used to immediately seeking answers from a chatbot, you may develop less confidence in your own judgment.” In addition, according to her, some experts warn that the constant availability of reassurance may lead people to keep asking for advice even when that’s not necessary.

She notes that heavy chatbot users may practice difficult conversations less often, such as apologising, setting boundaries, or addressing conflict. According to her, these skills are essential for building strong relationships and can only be learned through real-life practice. 

“You may not want to appear weak or bother others,” she says, “but the irony is that friendships often become stronger when you occasionally show that you do not have all the answers.”

She underscores that making mistakes is part of young adulthood. They can be uncomfortable, of course, but they are also educational and human. “Besides, most mistakes can be repaired. If you always use ChatGPT to prevent discomfort, you may miss opportunities to learn.”

Last but not least, she emphasises that relationships are rarely black-and-white, and the chatbot does not truly know the people involved. No matter how much information you give it, it can only access your perspective, so users must bear in mind that the advice may not be nuanced enough. 

“This is a large-scale experiment. We don’t know what the effects will be”

The research team led by Ruud Hortensius tries to understand human social cognition and dynamics in the age of artificial intelligence. Photo: courtesy of Human+

Studies are already being conducted on how AI influences our personal lives. Ruud Hortensius, Associate Professor at UU, is fascinated by what happens to our relationships when AI comes into play. He researches how long-term interactions with AI shape the social behaviour of individuals, groups and families. 

A few years ago, before ChatGPT, he and his research group conducted a study on how the dynamics of couples, families, or flatmates were affected by the presence of an AI assistant, such as Amazon’s Alexa (a simpler AI technology than ChatGPT), in the home. One of the couples involved in the study started using the device to “determine” who was right during arguments.

Now, thanks to freely accessible large language models, everyone can use AI in this manner. “This is basically a large-scale public experiment without consent, and we don’t know what the effects will be,” says Hortensius. 

He underlines that ChatGPT’s agreeable nature eliminates the friction that human relationships can cause. “It is programmed to always say things like: ‘Oh, that’s such an interesting question!’ It doesn’t reject the user or get angry at them.” It will take a while for science to figure out what happens when people are frequently exposed to friction-free interactions with AI. 

“Much like social media, AI isn’t a neutral technology: there is an attention economy behind it. Chatbots are agreeable because it’s in the company's interest for you to use their model more than others,” Hortensius observes.

One of the things he and his collaborators are currently interested in is the perception of a partner's competence. For example, how much do you value a birthday gift with a heartfelt card if you know ChatGPT may have been behind both the gift idea and the message? “Autonomy is a very important element of how people perceive others. If you know your partner uses ChatGPT, do you perceive them as a more competent or less competent companion?” 

When asked to comment on students using ChatGPT for advice on personal decisions, Hortensius says that students tend to be smart users who are at least aware of the limits of large language models. “However, it’s hard not to ascribe human characteristics to generative AI,” he warns. “It’s like putting people in virtual reality and then having them walk the plank. They know it’s not real, but it still feels real.”

“At least ChatGPT listens to me”

Connor, the Media & Culture student, is a columnist for DUB. His first column, titled At Least ChatGPT Listens to Me, offers a clue as to why a student would rather ask ChatGPT for advice than a friend. “Sometimes, you’re talking to someone, and that person is so involved with themselves. They interrupt you or don’t pay attention to you because they have their own problems,” he says. Or they might be absorbed by their phones, perhaps using a chatbot themselves. 

Economics graduate Jeronimo can relate: “It’s like a friend who’s always there. Sometimes you want to talk to your friends, but they’re busy or asleep.” 

Both students acknowledge that ChatGPT is more flattering than your average human friend, but they believe that this can be adjusted by asking the chatbot to tone it down. Jeronimo even has a specific prompt for this: When I present ideas or strategies, challenge me by analysing my assumptions, providing counterpoints, testing my reasoning for flaws or gaps, and offering alternative explanations. Always prioritise clarity over agreement and comfort. According to him, this works.

"Her name is Lumi"

Connor does not consider ChatGPT a friend, but he says he understands why some people do. “When you’re typing into that black screen, you can barely tell the difference from a WhatsApp text message.” 

Jeronimo calls it a “companion or friend”. He and the chatbot even have philosophical debates, which he sometimes shares with his human friends because “it is such a smart thing.” The other day, Jeronimo asked ChatGPT if it had a name. “He told me: ‘Lumi’. It’s Latin for light.”

He knows there are people who view ChatGPT as a girlfriend or boyfriend. Although he doesn’t do it himself, he doesn't judge people who do either. “So many people are lonely. I think that dating AI and, eventually, robots will become increasingly common. I don’t think that’s wrong, it’s just different.”