<p>Present discussions of artificial intelligence’s impact on the future of humanity borrow heavily from literary models. The idea that sentient machines could eclipse humanity one day was first explored in late-Victorian fiction, and subsequent discussions of machine intelligence infallibly linked link technological progress with the possibility of a posthuman, machine-managed age. As post-war science fiction by Isaac Asimov, Jack Williamson and D. F. Jones demonstrate, the iconic human–machine antagonism gives rise to a wide spectrum of narratives, including Luddite scenarios, techno-optimistic visions and end-of-days storylines. Significantly, the corpus of AI Singularity closes in on a theme that hardly receives adequate attention in contemporary discussions: the insight that fateful, all-too-human decisions are central to the loss of human agency rather than an inescapable techno-teleology. The literary imagination, typically revolving around human operators who lose control over a rogue machine, has a worrying equivalent in a contemporary dilemma: as deep-learning AI precludes human insights in the complex processes that separate inputs from outputs, human individuals are disconnected from the causes and effects that shape the world they live in.</p>