<p>This article develops a Lacanian analysis of conversational AI systems built on large language models by conceptualizing them as symbolic apparatuses that sustain desire rather than as neutral informational tools. Drawing on Jacques Lacan’s theory of object a, desire, transference, and the Big Other, the paper argues that conversational AI does not satisfy desire but structurally sustains it by translating desire into demand and responding without closure. Because large language models operate through the symbolic register while lacking subjectivity, they can occupy positions of authority, knowledge, and responsiveness traditionally associated with the Other, thereby intensifying repetition, attachment, and epistemic investment. The article further shows that affective responses such as anxiety and uncanniness emerge not only from AI failure but also from excessive symbolic proximity, when object a appears too directly. Situating these dynamics within the political economy of attention and engagement, the paper challenges dominant AI ethics frameworks centered on accuracy, transparency, and trust calibration. Instead, it proposes a critical shift toward examining how conversational AI reorganizes the symbolic conditions of desire, authority, and subjectivity. By foregrounding lack, limit, and symbolic distance, the article offers a theoretical framework for understanding why conversational AI is simultaneously compelling and destabilizing.</p>