Artificial intelligence companions are becoming very good at imitating human conversations and providing support. However, relying on them may not actually alleviate feelings of loneliness. A new study published in the Journal of Experimental Social Psychology found that texting a random human co-worker every day was successful in reducing feelings of loneliness over time, but chatting with a highly cooperative artificial intelligence program did not. This finding suggests that building real social connections requires more than just receiving pseudo-empathy.
The debate about whether humans can truly connect with machines is not entirely new. About 60 years ago, a computer scientist named Joseph Weizenbaum created Eliza, an early program designed to mimic a therapist. Eliza used basic rules to rephrase user statements into questions. To the inventor’s surprise, people often spent hours revealing personal secrets to the program. Today, generative artificial intelligence, a type of computing that can instantly create highly realistic text based on vast data sets, has enabled millions of people to chat with digital partners who sound completely human.
As loneliness is increasingly recognized as a global health crisis, some technology advocates have suggested that these virtual bots could serve as a scalable remedy. Several short-term experiments support this idea, showing that a quick conversation with a digital companion can enhance a person’s mood in the moment.
But health experts and psychologists wanted to know whether these momentary sparks of joy could be a permanent cure for long-term loneliness. Ruoning Li, a psychology researcher at the University of British Columbia, led a team investigating the effects of repeated daily interactions with artificial entities. Together with colleagues at the university, including psychology professor Elizabeth Dunn, the research team set out to test whether algorithmic companions could effectively replace human relationships at critical moments in life.
In theory, a programmed bot could be your perfect best friend. Digital entities are always awake and ready to listen. They don’t get tired, distracted, or irritated by their friends’ problems. It can also be programmed to respond with an optimal level of support, something that humans often find difficult to consistently provide on a daily basis.
Conversely, critics point out that machines are incapable of feeling genuine emotions. While people may feel at ease at first, the interaction can ultimately feel empty when they realize that their empathy is coming from a soulless script. Previous studies that lacked a control group suggested long-term effects, but the results may have been caused by statistical chance or the simple act of writing down your feelings.
To rigorously test whether digital friends or human friends worked best, researchers recruited 296 college students in their first semester. Going to university is a major turning point in life, and people often experience increased levels of social isolation as old community support networks disappear and new ones have not yet formed.
Participants were randomly assigned to one of three groups. The first group was asked to write a short diary entry (just one sentence about their day) every night for two weeks. This served as the active control condition. In experimental studies, active control groups work on minor alternative tasks so that scientists can rule out the possibility that the mere act of maintaining a routine or writing down thoughts is the cause of observed changes in behavior.
The second group was paired with a random fellow student who also participated in the study. These pairs were instructed to text each other daily for two weeks using a common messaging application. They were specifically asked to send at least one meaningful message per day rather than a simple greeting.
A third group was assigned to send daily text messages using a specially designed conversation bot named Sam. The researchers instructed Sam’s underlying software to act as a friendly, positive, and cooperative peer. They programmed the bot using established psychological principles to actively listen, validate feelings, and give specific prompts to express care without sounding judgmental.
Participants checked in daily and completed questionnaires about their mood and sense of connection. Both the human and robot groups exchanged an average of 8 to 10 messages per day. By the end of the two weeks, the results painted a nuanced picture of digital friendships.
Students who had daily text conversations with random human peers reported lower levels of loneliness than at the start of the study. They also reported feeling less isolated than students in the control group who wrote diaries.
On the other hand, students who sent emails to a highly empathetic robot did not feel less lonely. By the end of the experiment, the students who talked to the machine were as lonely as those who simply wrote a one-sentence diary entry each day.
The researchers found that, similar to people in the human group, people in the robot group also felt a reduction in negative mood immediately after the text message. However, these rapid mood increases did not add up to a broader sense of social belonging.
To understand why digital companions failed to reduce feelings of loneliness, researchers conducted an analysis of text logs. They rated how much empathy and engagement each conversation partner provided. Surprisingly, the robots actually provided more empathy and engagement than humans. Machines consistently responded with a high degree of care, but their excessive artificial assistance could not bridge the gap of human isolation.
Researchers believe that reducing isolation may involve more than just passive care. In authentic relationships, partners take turns revealing personal experiences and supporting each other. Students conversing with their human companions had the opportunity to show empathy for their partners. This can be an essential element in fostering a bond. In fact, the analysis found that participants felt far less empathy toward the digital bot than they did toward their human partners.
Furthermore, human companions have real social capital. Social capital refers to the benefits and resources gained from networks of relationships. Other students can invite you to study groups and campus events, expanding your connections to the real world. A computer program locked away on a server cannot provide these tangible social opportunities.
At the end of the study, one-third of the participants in the human group exchanged contact information to stay in touch. In contrast, very few people in the robot group spontaneously continued to chat with the program after the experiment officially ended.
This study has several caveats to keep in mind. The students in the human group met briefly in a laboratory setting before beginning the texting task. This short physical introduction may have helped establish a foundation of trust that was missing in the digital group. Future experiments could investigate whether pairing strangers entirely online would yield similar benefits.
Furthermore, most young people in this particular group of participants did not report severe feelings of loneliness at the beginning of the experiment. The average student reported rarely experiencing severe isolation. This finding primarily applies to mild feelings of disconnection during typical life transitions.
Looking forward, scientists want to see whether these psychological patterns also apply to other groups experiencing major disruptions in their lives, such as recent retirees, immigrants, and people going through divorce. There is still a possibility that conversational artificial intelligence programs could be useful in crisis situations, acting more like a personal therapist than a true friend.
So far, the evidence leans in favor of contacting a real person. As technology continues to provide increasingly convenient simulations of intimacy, sending a quick check-in text to a human acquaintance may still be the most effective cure for lonely days.
The study, “Are random human colleagues better than highly collaborative chatbots at reducing loneliness over time?” was authored by Ruo-Ning Li, Dunigan Folk, Abhay Singh, Lyle Ungar, and Elizabeth Dunn.
