Patient use of AI and its implications for transference–countertransference dynamics

Artificial intelligence (AI) is appearing in the analytic space – as a topic but, at times, as something that can become part of the patient's psychic reality. Analysts may need to adopt a sustained, clinically grounded stance toward this phenomenon – one that recognizes initial defensive reactions and preconceptions, yet is not determined by them. Drawing on clinical observations, it explores how patients' use of large language models as digital confidants can shape transference: from easing the articulation of thoughts behind the “second censorship,” to producing insights that are dissociated from affect. The focus is not on whether AI produces valid interpretations, but on how the analyst receives and metabolizes AI-mediated communications within the transference–countertransference field. Feelings such as rivalry, irritation or protectiveness may offer valuable analytic data, reflecting both the analyst's vulnerabilities and the patient's unconscious enactments. In some treatments, AI may be positioned as a third figure in the analytic dyad, evoking triangular tensions and bringing conflicts to the surface. These moments can provide clinically relevant information about internal object relations. Such material should be taken seriously and engaged with as part of the analytic process, rather than being avoided or downplayed – consciously or unconsciously – by the analyst.