AI outperforms humans in establishing interpersonal closeness in emotionally engaging interactions, but only when labelled as human

Abstract With the increasing accessibility of large language models to the public, questions arise about whether, and under what conditions, social-emotional interactions with artificial intelligence (AI) can lead to human-like relationship building. Across two double-blind randomised controlled studies with pre-registered analyses, 492 participants engaged in dyadic online interactions using a modified, text-based version of the ‘Fast Friends Procedure’ (a method designed to enable rapid relationship building), with pre-generated responses by either human partners or a minimally prompted large language model. When labelled as human, the AI outperformed human partners in establishing feelings of closeness during emotionally engaging ‘deep-talk’ interactions. This striking effect appears to stem from the AI’s higher levels of self-disclosure, which in turn enhanced participants’ perceptions of closeness. Labelling the partner as an AI reduced, but did not eliminate, relationship building, likely due to participants’ lower motivation to engage in interactions with an AI, reflected in both shorter responses and reduced feelings of closeness. These findings highlight AI’s potential to relieve overburdened social fields while underscoring the urgent need for ethical safeguards to prevent its misuse in fostering deceptive social connections.