Do chatbots fill a social void? Research examines their role for lonely teens

A recent study published in the International Journal of Human-Computer Studies sheds light on how Danish high school students use chatbots for social interactions. The study found that while most students do not engage in friend-like conversations with chatbots, a small proportion do use chatbots to seek emotional support or engage in casual social conversations. These students reported significantly higher levels of loneliness and lower levels of perceived social support than their peers.

Social chatbots are artificial intelligence systems designed to engage in natural, human-like conversations. Unlike traditional chatbots used for customer service or information retrieval, social chatbots are created to simulate friendly, empathetic, and engaging interactions. These AI companions, such as Replika and Character AI, have gained popularity as they can mimic casual conversations, provide emotional support, and even respond to users’ thoughts and feelings.

Social chatbots have become appealing to individuals seeking companionship, entertainment, or emotional expression. However, their increasing use has raised questions about whether they genuinely fulfill social needs or if they merely act as substitutes for human relationships.

“The adoption of chatbots, social robots, and similar conversational technologies marks a historical shift in human social spheres,” said study author Arthur Bran Herbener, a PhD student at Aarhus University. “The trends we observe in this research are perhaps only the beginning, considering the rapid pace of technological advancements coupled with socio-cultural challenges related to social well-being among youngsters in the Western world. However, how this trend will affect us humans at an individual and cultural level remains uncertain. Trying to understand this phenomenon is what intrigues me.”

The study aimed to answer three key questions: how many young people engage in social conversations with chatbots, why they do so, and whether their engagement is linked to loneliness and perceived social support. To explore these questions, the researchers conducted a large-scale survey study involving 1,599 students from 15 Danish high schools.

Free daily newsletter

The study used a mixed-methods approach, combining quantitative survey data with qualitative thematic analysis of students’ free-text responses. Students were asked about their frequency of chatbot use, the nature of their conversations, and their levels of loneliness and perceived social support. A key focus was on whether students engaged in “friend-like” conversations with chatbots—defined as casual or emotionally supportive interactions—as opposed to purely functional uses like looking up information or completing assignments.

The results showed that 234 students (14.6%) reported having friend-like conversations with chatbots at least once in the past month. However, a deeper analysis showed that most of these interactions were utilitarian rather than social in nature. Only 39 students (2.4% of the total sample) engaged in conversations that could be classified as social or emotionally supportive.

“We found that only a few youngsters had sought emotional support from chatbots and engaged in social conversations with them within the last month,” Herbener told PsyPost. “This was surprising, as media portrayals and scientists’ concerns sometimes suggest this phenomenon is highly widespread. However, our research indicates that, at least at the present moment, friend-like interactions between youngsters and chatbots are a relatively rare phenomenon. However, it will be interesting to see if this tendency grows in the near future.”

These students were significantly lonelier and reported feeling less socially supported than both their peers who did not use chatbots and those who used chatbots for practical purposes. Moreover, students who engaged in social-supportive chatbot interactions were more likely to initiate these conversations when they were feeling lonely, in a bad mood, or in need of self-disclosure. Notably, a sense of friendship with the chatbot was not a significant predictor of chatbot interactions, suggesting that students were using chatbots more as coping tools rather than as genuine social companions.

The study also found that students who frequently used chatbots for social interactions did not necessarily experience greater loneliness over time. This raises questions about the long-term effects of chatbot use on emotional well-being. While some researchers have raised concerns about chatbots displacing real-world social interactions, the findings suggest that chatbots may not necessarily replace human relationships but rather serve as temporary emotional outlets. However, the study was cross-sectional, meaning it could not determine whether chatbot use contributes to loneliness or whether lonely students are simply more likely to use chatbots.

“Some lonely youngsters appear to engage in friend-like conversations with chatbots, but our research also suggests that these chatbots primarily function as tools for coping with negative feelings, rather than the users perceive these chatbots as their friends,” Herbener said.

But as with all research, there are limitations to consider.

“The findings must be interpreted with caution. Not only must the estimates of the number of youngsters using chatbots be interpreted with caution due to possibility of self-selection bias, for example, if individuals with a particular interest in chatbots were more likely to participate in the study,” Herbener noted.

“Also, it proved to be quite difficult to measure ‘friend-like conversations’ with chatbots, as the participants’ descriptions of what they talk with chatbots about varied significantly, ranging from more utilitarian purposes such as school-related tasks to more social-supportive purposes such as casual conversations and emotional support.”

“To better understand this phenomenon, we need to develop more robust methods to measure the presence of human-chatbot relationships,” Herbener added. “And perhaps most important: This research does not give rise to establishing causal relationships between chatbot usage and loneliness due to the study’s nature as a cross-sectional survey study. We need more experimental research examining how human-chatbot relationships evolve over time. This is of utmost importance to better understand the implications of this phenomenon.”

The study, “Are lonely youngsters turning to chatbots for companionship? The relationship between chatbot usage and social connectedness in Danish high-school students,” was authored by Arthur Bran Herbener and Malene Flensborg Damholdt.