Moral imagination is the specific cognitive capacity that cruelty by default requires to operate. It is not benevolence, not compassion, not the general disposition to care about others. It is the concrete cognitive act of projecting forward from a decision in the present to the lives of specific people who will be affected by it, none of whom are in the room, most of whom the decision-maker will never meet. The capacity is not innate. It is cultivated through specific practices and requires specific conditions — time to reflect, time to consult, time to imagine the person downstream. These conditions are the first casualties of environments optimized for speed, and their absence is the primary mechanism through which avoidable cruelty becomes the predictable output of intelligent, well-intentioned builders.
The capacity was an analytical preoccupation of moral philosophers from Adam Smith's impartial spectator through Iris Murdoch's sustained attention to the reality of other persons. Shklar inherits this tradition and sharpens it through her structural analysis. The capacity is not primarily a virtue of character. It is a function of institutional environment. The builder who works within deployment cycles measured in days cannot exercise moral imagination at a meaningful depth, not because he lacks character but because the time required for the cognitive operation has been structurally eliminated. The conditions for moral imagination — the unoccupied intervals during which the imagination can operate — have been colonized by productive activity.
The AI transition intensifies the destruction of these conditions in specific ways. The Berkeley researchers documented task seepage — the colonization of previously protected temporal spaces by AI-accelerated work. Lunch breaks filled with prompts. Waiting rooms converted to workstations. The cognitive gaps between tasks eliminated by the tool's constant availability. Each of these colonized intervals was previously a potential site of moral imagination — a space in which the builder might have paused to consider the person downstream. The colonization is not incidental. It is produced by the same tools whose downstream effects the colonized intervals would have otherwise permitted the builder to consider.
The political significance of the capacity's atrophy is that it produces a population of builders who are structurally incapable of exercising the moral agency their positions demand. This is not a moral indictment. It is a structural diagnosis. The builder who ships a product that intensifies work without adequate consent has not failed as a person. He has operated within an institutional environment that has systematically prevented the exercise of the capacity he would have needed to recognize the problem. The responsibility, in Shklar's framework, migrates upward. It shifts from the individual builder to the institutional environment that prevented individual moral agency from operating.
The institutional remedies the framework demands are therefore specific. Deployment intervals that create time between the decision to ship and the act of shipping. Mandatory impact assessments that require the articulation of foreseeable consequences before consequences become unrecoverable. Professional norms that treat the question "should we?" as a mark of seriousness rather than a failure of nerve. Protected time for the cognitive operations that moral imagination requires, treated as a productivity input rather than a productivity cost. These interventions do not prohibit building. They create the conditions under which building can occur with the moral attention that technological power at this scale demands.
The term has deep roots in moral philosophy — Edmund Burke, Adam Smith, Iris Murdoch — and received significant contemporary development in Martha Nussbaum's work on narrative imagination. The Shklar volume synthesizes these threads with Shklar's specific structural analysis of how institutional conditions shape moral capacity.
Imagination is cognitively expensive. The act of projecting forward to the lives of specific affected persons is not instantaneous; it requires time that environments optimized for speed do not provide.
Conditions are institutional, not characterological. Whether builders can exercise moral imagination depends on the environment they work within, not primarily on their individual dispositions.
AI deployment eliminates the conditions. Task seepage colonizes precisely the intervals during which moral imagination could have operated.
Atrophy is structural, not moral. The builder whose moral imagination has atrophied has not failed as a person; the institutional environment has failed the person.
Remedies require protected time. Deployment intervals, impact assessments, and professional norms that protect reflective space are the specific institutional structures the capacity requires.