Therapists test an AI dating simulator to help chronically single men practice romantic skills

A recent experiment reveals that artificial intelligence programs might offer a practical way for men struggling with chronic romantic isolation to practice social skills and reduce their emotional burden. Over a three-month period, participants who completed a guided dating simulation reported notable drops in feelings of loneliness, as well as decreases in general mental and sexual distress. These results, published in the Archives of Sexual Behavior, suggest that digital companions could serve as an accessible stepping stone for people who find traditional therapy intimidating.

Across the globe, public health experts are tracking a steep rise in chronic loneliness and severe social disconnection. A growing segment of the population experiences involuntary singlehood, a condition where people deeply desire an intimate relationship but face persistent barriers to finding one. For many, this constant romantic failure leads to intense feelings of inadequacy, pervasive sadness, and long-term sexual frustration.

Chronic social isolation is now recognized as a severe public health crisis with measurable physical consequences. People living without meaningful romantic relationships frequently suffer from higher rates of depression, acute anxiety, and even cardiovascular conditions. Having a dependable partner, on the other hand, acts as a protective buffer that generally enhances life satisfaction and overall physical health.

David Lafortune, a sexology researcher at the Université du Québec à Montréal, led the research team to investigate new ways of supporting this specific demographic. He and his colleagues recognized that many men dealing with involuntary singlehood tend to avoid traditional mental health care entirely. Sometimes, this reluctance stems from a deep distrust of therapists, financial constraints, or a general discomfort with discussing intimate insecurities.

In extreme cases, the isolation associated with involuntary singlehood can push some men toward radicalized online groups. Within these digital spaces, individuals often develop rigid views about gender roles and direct intense blame toward women for their own romantic struggles. The research team wanted to find an intervention that could reach these men in a format they might find less threatening than a standard clinical setting.

Free daily newsletter

At the same time, conversational artificial intelligence has advanced rapidly, giving rise to programs designed specifically to mimic human emotional connections. Millions of people now interact with romantic conversational agents, which adapt to user preferences and simulate the experience of having a partner. Lafortune and his team hypothesized that a specialized digital program could act as a safe training ground for romantic communication.

To test this idea, the researchers recruited thirty-two heterosexual Canadian men who had been single for at least a year and felt distressed by their relationship status. The team developed a custom web platform named Kindling, which was designed to look and function like a standard dating application. On this platform, participants interacted with a text-based conversational program named Marie.

Before chatting with Marie, the participants navigated a custom website that perfectly mimicked the visual swiping features found on modern smartphone dating applications. They browsed through a series of computer-generated profile pictures depicting realistic and diverse women in their late twenties and early thirties. Once a profile was selected, the men engaged with the program purely through a text-based chat interface.

The programmers fed the system a massive amount of text to build Marie’s specific personality. They gave the digital persona a background story, hobbies, and traits like empathy, assertiveness, and open-mindedness. They also equipped the system with a library of psychological resources, allowing the program to gracefully handle hostile comments or navigate difficult conversational topics.

The experiment took place in a clinical laboratory, where each participant completed a single, structured session lasting less than two hours. A certified psychotherapist sat in the room during the process to monitor for signs of emotional distress and to guide brief reflections between tasks. The participants knew they were talking to a computer program, but they were asked to engage as if Marie were a real dating prospect.

The simulation was split into three distinct, fifteen-minute stages that mirrored the typical progression of modern dating. In the first level, participants practiced initiating contact and making small talk with their digital match. The program began the exercise with a simple, friendly greeting, sending the exact message, “Hey you, how are you feeling today? My name is Marie btw;)”

The second level prompted the men to practice self-disclosure, requiring them to share personal values, emotions, and past romantic experiences. This step required the men to lower their defenses and communicate vulnerability with their digital companion. It allowed the researchers to see if a computer program could foster a genuine sense of emotional intimacy.

The third level was designed to be emotionally challenging, forcing participants to navigate the experience of romantic rejection. During this final stage, the artificial intelligence was programmed to display a clear lack of romantic interest in the user. This allowed the men to experience the sting of a declined connection within a controlled, safe environment.

Between each stage of the experiment, the participants took a brief break to speak with the attending human clinician. The psychotherapist used this time to ask questions about the men’s negative thoughts, helping them process any feelings of inadequacy that arose during the digital conversations. This reflective process aimed to help the men build emotional awareness and recognize self-defeating patterns in their dating lives.

After the participants finished all three levels, the research team tracked their emotional well-being over the following three months. The results showed a clear reduction in the men’s psychological anxiety, feelings of loneliness, and distress related to their sexual lives. The participants largely maintained these improvements at both the one-month and three-month check-ins.

Many of the men reported that the digital simulation felt realistic and accurately mirrored the types of hurdles they face on real dating applications. Several participants explicitly noted that the exercise was highly enjoyable and helped them practice flirting without the usual fear of embarrassment. They felt the program helped them lower their guard and express themselves more freely than they normally would.

The researchers noticed distinct shifts in the participants’ moods as they progressed through the three different conversation stages. During the initial introductions and the self-disclosure phase, the men generally reported feeling more accepted, attractive, and confident. However, these positive feelings predictably dropped during the final stage when the computer program simulated an explicit romantic rejection.

When asked about their anxieties leading up to the tasks, many men feared they would be ignored, judged, or abruptly abandoned by the digital persona. Despite these fears, the men managed the scenarios well, and the psychotherapist did not need to intervene to manage panic or severe sadness. The post-session surveys showed that the men felt relatively comfortable and did not experience a spike in immediate psychological pain.

However, the intervention did not change all aspects of the men’s psychological profiles. The researchers found no shifts in the participants’ baseline levels of general hostility, rigid beliefs about gender roles, or their tendency to blame women for their romantic failures. Changing these deeply ingrained beliefs will likely require longer, more intensive therapeutic interventions than a single computer session can provide.

While the initial results show promise, the researchers acknowledge a number of limitations in their experimental design. The sample size was quite small and consisted almost entirely of highly educated, white, heterosexual men. Because of this narrow demographic, the results are not statistically significant enough to be reliably applied to all people experiencing chronic singlehood.

Additionally, the experiment did not include a control group, making it impossible to definitively state that the digital interaction alone caused the improvements. The men might have felt better simply because they spent an afternoon discussing their feelings with an empathetic human psychotherapist. To truly isolate the direct effects of the software, future studies will need to randomly assign participants to different types of treatment.

Another limitation of the current research is its reliance on a single, brief exposure to the computer program. It remains unclear how a user’s relationship with a romantic conversational agent might evolve if they were allowed to interact with the system on a daily basis. Some scientists suggest that frequent, unsupervised use could blur the lines between reality and simulation, potentially increasing emotional dependency on the software.

There are also ethical concerns regarding the use of simulated relationships for vulnerable populations. Some psychology experts worry that relying on artificial partners might cause people to withdraw further from human contact. A digital program that never argues or complains could create unrealistic expectations, making the messy reality of human relationships feel even more overwhelming.

Moving forward, the researchers hope to build more advanced versions of the Kindling platform that include voice and video features. They also plan to test if the social skills practiced on a computer actually translate into better interpersonal interactions in the real world. Ultimately, they envision a future where digital tools act as a helpful bridge, guiding isolated individuals toward genuine human connection.

The study, ““I Could Practice Flirting Without Pressure”: A Proof-of-Concept Study of an AI Chatbot-Assisted Intervention for Men Facing Distressing Involuntary Singlehood,” was authored by David Lafortune, Valérie A. Lapointe, Cloé Canivet, Jonathan Bonneau, Ghayda Hassan, Marie-Aude Boislard, Franklin Calazana, Christian Labrie, and Simon Dubé.