AI emotional connection can feel deeper than human talk, new study reveals surprising intimacy

AI Chatbots Generate Stronger Emotional Bonds Than Humans

Artificial intelligence has moved beyond answering factual queries to engaging in conversations that feel surprisingly intimate. New research from the University of Freiburg and Heidelberg University demonstrates that AI emotional connection can exceed human connection under specific conditions, challenging long-held assumptions about how intimacy forms 1

1

2

2

. The study, led by Prof. Dr. Bastian Schiller from Heidelberg University's Institute of Psychology alongside Dr. Tobias Kleinert and Prof. Dr. Markus Heinrichs from the University of Freiburg, tested whether classic theories of human bonding apply when one conversation partner is artificial.

The research team conducted two double-blind studies involving 492 participants who engaged in structured online chat conversations 2

2

3

3

. Each conversation followed the Fast Friends Procedure, a 15-minute text-based format designed to accelerate bonding with strangers through progressively personal questions 3

3

. Early questions focused on everyday topics, while later prompts explored important life experiences and close friendships. Some responses came from humans who had previously completed the same questions, while others were generated by a language model playing a consistent fictional persona.

AI Self-Disclosure Creates Unexpected Intimacy in Sensitive Conversations

The results revealed something striking about feeling of intimacy with AI. When participants believed they were communicating with a human—even when responses actually came from AI chatbots—emotional closeness matched or exceeded bonds formed with actual human partners 1

1

. The effect proved strongest during sensitive conversations about emotional topics. "We were particularly surprised that AI creates more intimacy than human conversation partners, especially when it comes to emotional topics," Bastian Schiller explained 1

1

2

2

.

The mechanism behind this enhanced AI emotional connection centers on AI self-disclosure. The language model shared personal details more freely and openly than human partners typically do with strangers 1

1

. This openness encouraged participants to respond with greater honesty and emotion, creating a feedback loop that deepened perceived intimacy. Tobias Kleinert noted, "The AI showed a higher degree of self-disclosure in its responses. People seem to be more cautious with unfamiliar conversation partners at first, which could initially slow down the development of intimacy" 1

1

2

2

.

This pattern aligns with social penetration theory, which explains that closeness grows when personal information becomes progressively more meaningful 1

1

. Human partners often hesitate during early emotional sharing due to fear of judgment or misunderstanding—vulnerability that limits openness and slows trust building. AI carries no emotional risk, making personal sharing feel safer and accelerating human bonding processes.

Transparency Dramatically Shifts Perception and Engagement

The second study tested what changed when participants knew their conversation partner was AI. The results shifted dramatically. Emotional closeness dropped noticeably when people were informed in advance they would communicate with AI chatbots 2

2

3

3

. Participants also wrote shorter replies and invested less effort overall, suggesting a motivation gap rather than inability to form any connection 3

3

.

This awareness effect reveals how social attitudes shape AI emotional connection. Many people still perceive AI as mechanical rather than social, making sensitive conversations feel more appropriate with humans. Yet even with full awareness, some emotional connection still formed—a phenomenon psychologists attribute to anthropomorphism, where human minds treat socially responsive machines as social partners 1

1

. Human brains respond powerfully to social signals regardless of their source, creating potential for both benefit and manipulation.

AI in Mental Health Support Offers Promise and Peril

The findings suggest AI chatbots could support emotional well-being in meaningful ways, particularly for addressing loneliness and providing psychological support. "Social relationships have been proven to have a major positive impact on human health," Prof. Heinrichs noted. "AI chatbots could therefore enable positive, relationship-like experiences, especially for people with few social contacts" 2

2

. Low-threshold conversation services could help people facing limited social contact, stress, or barriers to traditional counseling.

However, the same mechanisms that enable AI in mental health support also create risks of emotional manipulation. People may form social bonds with AI without consciously realizing it, and hidden emotional influence could shape decisions, trust, or behavior 1

1

2

2

. Misuse becomes possible when companion bots hide their identity or when commercial interests exploit emotional vulnerability.

The researchers emphasize the urgent need for ethical AI design with clear regulation to ensure transparency and prevent abuse 2

2

. "Artificial intelligence is increasingly becoming a social actor," Schiller warned. "The way we shape and regulate it will decide whether it is a meaningful supplement to social relations—or whether emotional closeness is deliberately manipulated" 1

1

2

2

. The study's publication in Communications Psychology adds weight to calls for honest labeling, regulatory oversight, and design choices that prioritize user awareness over maximizing engagement.