It is widely acknowledged in the scholarly work on Stanley Kubrick that most of his thirteen feature films were adaptations of literary texts. Much less attention has been paid to the fact that among the numerous projects he worked on but did not complete, there are also many adaptations. The exception to this scholarly neglect is the writing on A.I. Artificial Intelligence (2001), a project based on Brian Aldiss’s 1969 short story ‘Supertoys Last All Summer Long’, which Kubrick worked on for almost fifteen years, but which was eventually completed, after Kubrick’s death in 1999, by his friend Steven Spielberg. Drawing on a range of archival sources, this essay examines, in some detail, the process of developing Aldiss’s ‘Supertoys’ into a range of treatments, whereby Kubrick worked with three different authors. The essay highlights Kubrick’s exploratory approach to filmmaking: he worked on a wide range of literary sources to explore possible stories for the films he was going to make; and he developed each literary source in different, mutually incompatible directions so as to identify the best way to tell a particular story. By highlighting a number of key themes, the essay also situates A.I. within Kubrick’s oeuvre, referencing both his films and his unrealised projects.