With DeepSeek (the newest competitor to ChatGPT) making the headlines, and with the season of ~love~ upon us, some are pulling a Ryan Gosling in Blade Runner 2049 and getting AI girlfriends to share our nights with. But are there negative consequences for this, and how could it change our personal relationships in future?  

A chatbot is a type of AI you chat to. While models such as ChatGPT technically count, there are others which perform specifically to provide connection, such as Replika (30 million users) and Character AI (28 million active monthly users). While typical users engage with little consequence, some have darker experiences that effect their lives. In 2021, an AI girlfriend urged a teen boy to kill the Queen. In 2023, another bot directly encouraged a Belgian man to take his own life. 

On the other hand, some people use these services to enhance real-life relationships, rather than replace them. One autistic woman told Concrete: 

 “I use AI to help write difficult messages… it helps me know if what I’m saying makes sense…won’t judge what I say … It validates what I write and makes me feel listened to.”

While their use as therapeutic tools has been praised by clinical psychologists, it always comes with the caveat that chatbots should not be used to replace human interaction, the highlighted issue often coming around to their lack of true empathy.  

Studies show bots are surprisingly good at identifying human emotion, though responses “lack true emotional depth”. Bots tend to mirror what is said to them back to the user which, while validating, prevents further discussion and can lead to avoidance of difficult feelings – long known by psychologists to only enforce them. One psychologist used the analogy of using opiates instead of performing physiotherapy: one is hard but is a solution, one merely masks the problem.  

Combined with the ease of interaction with AI as opposed to their human counterparts (24/7 access and fast response times), use can set an unrealistic expectation for real-world relationships and can cause an increased sensitivity to rejection. 

MIT researcher Sherry Turkle notes, like social media, a decreased sense of vulnerability. This corresponds to a lower sense of intimacy, and, thereafter, a lower sense of empathy. In the same way you are less intimate with your professor than your friends, you are more likely to bad mouth them while sitting in Blue Bar, less likely to think about their human tendencies to feel or make mistakes.  

Also shared with social media are tendencies of users to addiction or over-reliance, however with much different effects on users mental health. Though not much long-term research has been conducted, a not insignificant number of cases found chatbot use in teens to reduce symptoms of anxiety and depression, with Psychology Today labelling chatbots “a significant advancement in mental health support”.  

With a lack of substantial research into the subject, it’s hard to tell what the long term effects of chatbots on our social and emotional future. An immense stride in the personalisation and availability of mental health treatment? The beginning of (or slight further shuffle to) the end of genuine human connection and fulfilling relationships? One thing that remains ever constant, poignant, and relevant is this:

“Technology is neither good nor bad: nor is it neutral.”