In mainstream public debates, AI is consistently measured against a presumed human baseline. Because this comparative register privileges a one-dimensional definition, "the human" now functions as the medium through which AI becomes thinkable, lovable, governable, and saleable. Yet the gap between AI's uneven capabilities and the intensity of public reactions (panic, awe, hope, or resignation) suggests that our projections of AI's future originate less in the technology itself than in our imagination of ourselves. To theorize this dynamic, I propose a definition of AI as an extimate technology. AI externalizes our most intimate cultural artifacts and returns them to us in a form that appears to belong to someone else. I show that technologists and futurists who anticipate the technological singularity fail to see this constitutive relation and are thus unable to ask questions that could guide a path toward a more ethical future of AI.