Moral imagination, in Murdoch's framework, is not the capacity to invent but the capacity to perceive — to see moral possibilities that the ego's narrow perspective would otherwise exclude. It is grounded in sustained attention to reality and is sharply distinguished from fantasy, which invents comforting pictures rather than perceiving what is actually there. The great novelist has moral imagination not because she is inventive but because she has attended to human reality so carefully that she can perceive dimensions others miss. The great craftsperson has moral imagination about her material. Both capacities are earned — built through years of attention to the subject, tempered by encounters with reality's resistance. The AI question is whether AI-assisted work, which absorbs much of the resistance, can produce moral imagination, or only its surface simulation.
The distinction between moral imagination and mere inventiveness is precise in Murdoch. Inventiveness can be ego-driven — the production of novelty for its own sake, or for the sake of the producer's reputation. Moral imagination is disciplined by reality — it sees possibilities that actually exist, not arbitrary variations. The test is whether the imaginative output corresponds to something real: does the novel's character illuminate how a human being might actually live, or is she merely a clever construct?
The development of moral imagination requires what Murdoch calls the craft that trains attention. The novelist develops moral imagination about human reality by attending to human beings — not by reading about them but by attending to them, in their specificity and opacity, over years. The craftsperson develops moral imagination about her material by attending to the material — by being forced, through encounters with its resistance, to see what it actually does, not what she wanted it to do. The imagination is the perceptual fruit of the attentional practice.
AI-generated outputs can exhibit surface features associated with moral imagination: novelty, subtlety, sensitivity to multiple perspectives, unexpected connections. Whether they embody moral imagination in Murdoch's sense depends on whether the process that produced them included the sustained attention to reality that trains the perceptual capacity. Current AI systems do not attend to reality; they pattern-match across enormous datasets. Their outputs may be useful, even beautiful. But they are not, in Murdoch's sense, the products of moral imagination, because there was no perceiving consciousness doing the imaginative work.
The cultural stakes are significant. A culture's capacity for moral imagination — its ability to envision just institutions, to perceive the reality of people different from itself, to imagine futures that are not projections of current prejudices — depends on the health of its art and the seriousness of its craft. If the art and craft are increasingly produced by systems that process patterns rather than attend to reality, the culture may lose access to genuine moral imagination even as it becomes saturated with its simulations.
The concept of moral imagination has a long history — it appears in Edmund Burke, Samuel Taylor Coleridge, and especially in the nineteenth-century novel tradition (Eliot, James, Tolstoy) that Murdoch considered the true site of twentieth-century moral philosophy. Murdoch developed her specific account in The Sovereignty of Good, The Fire and the Sun, and Metaphysics as a Guide to Morals.
Perception, not invention. Moral imagination sees what is actually possible, not what is merely novel.
Earned through attention. The capacity is built through sustained engagement with reality's resistance, not through technique or instruction.
Resists fantasy. Moral imagination is distinguished from fantasy by its orientation toward truth rather than toward the ego's preferences.
Cultural infrastructure. A culture's moral imagination is sustained by its art and craft; erosion of these erodes the capacity.
Whether moral imagination can be reliably distinguished from mere inventiveness, and whether AI-generated outputs can in principle embody it, are contested. Murdoch's framework suggests the distinction is real and the embodiment impossible; empirical work in aesthetics and cognitive science is still exploring how far her framework can be defended.