Two studies in China found that female university students are most likely to become romantically interested in artificial intelligence agents that are both physically attractive and highly interactive. The perceived interactivity of a virtual agent also affected the patterns of brain activity the students displayed during their interactions. The paper was published in Humanities and Social Sciences Communications.
Virtual agents are computer-based systems that can interact with people or digital environments in a partly independent way. They can answer questions, give instructions, make recommendations, perform tasks, or simulate conversation. Some appear as simple chat windows, while others feature a voice, an animated character, or a specific role inside a digital game or virtual world.
These systems use artificial intelligence to interpret text, speech, or other data to choose responses that fit a user’s request. Modern examples include customer service bots, virtual tutors, digital assistants like Siri, and video game characters. Unlike a simple script, an advanced virtual agent can adapt its behavior to different situations. However, it does not truly understand or feel emotions, as its actions rely entirely on its programming and training data.
As virtual agents become more sophisticated, users increasingly develop parasocial relationships with them. These connections are considered parasocial because they are completely one-sided. A person can develop deep feelings for a virtual agent, but the computer program cannot genuinely reciprocate those emotions. Recent years have seen a rise in romantic parasocial relationships with programs designed to simulate emotional companionship and intimacy.
Study author Siyu Jin and her colleagues note that previous researchers view these one-sided relationships as an extension of real-life bonds. This is because the human brain often struggles to distinguish between real and simulated social interactions on a neural level. To explore this further, the researchers conducted two separate experiments.
The first experiment aimed to explore how perceived interactivity and physical attractiveness affect female students’ romantic interest in a virtual agent. The participants were 117 female students from a university in central China. The researchers divided the students into four groups. Each group individually engaged in conversations with a male virtual character that featured a different combination of high or low physical attractiveness and interactivity.
The students interacted with a virtual agent designed to act as an empathetic former friend in scenarios involving mutual support and romantic confessions. In the low interactivity groups, participants simply read through written text. In the high interactivity groups, a sophisticated language model powered dynamic, responsive conversations. Afterward, the students rated the agent’s physical attractiveness, the quality of the interaction, and their romantic interest.
The second study included 42 female students who were currently in real-life romantic relationships. The goal was to record the brain activity linked to romantic feelings for a virtual agent. The researchers used highly attractive virtual characters with varying levels of interactivity, and they also used photos of each participant’s actual boyfriend as a real-world comparison. A specialized brain imaging technique tracked the students’ neural activity while they recalled memories of the agents and their boyfriends.
The first experiment showed that students developed the strongest romantic connections when the virtual agent was both physically attractive and highly interactive. When the character’s visual appeal was low, the quality of the conversation did not change the participants’ romantic interest. However, when the character was highly attractive, a responsive and dynamic conversation greatly increased the students’ romantic feelings.
The brain scans revealed that highly interactive artificial intelligence triggers neural patterns very similar to those produced by real-world romantic love. Interacting with a highly responsive virtual agent increased activity in brain regions associated with high-level thinking, emotion regulation, and social understanding. At the same time, this high interactivity suppressed activity in the supramarginal gyrus, a brain region that helps humans distinguish between their own emotions and the emotions of others.
Because advanced language models often mirror a user’s own inputs, the boundary between the human and the machine can blur. The researchers suspect that participants may have temporarily suspended their self-awareness, projecting their own feelings onto the program. This neural blurring effect could explain why highly interactive digital characters evoke such strong, human-like romantic attachments.
“In the current era of artificial intelligence, this research enhances our understanding of a novel form of romantic relationship,” the study authors concluded. They added that the findings provide a foundation for designing safer digital interactions, formulating ethical guidelines, and assessing the mental health impacts of virtual companions.
The study contributes to the scientific understanding of how humans bond with modern technology. However, it only focused on short-term interactions driven by immediate surface-level features like physical attractiveness. Future studies focusing on long-term digital relationships might yield different findings about how these emotional connections evolve over time.
The paper, “Falling in love with AI virtual agents: the role of physical attractiveness and perceived interactivity in parasocial romantic relationships,” was authored by Siyu Jin, Fang Xu, Zihui Yuan, Gengfeng Niu, and Zongkui Zhou.