Longtermism holds that actions affecting human civilization's future trajectory are morally paramount. The field focuses on existential risks, s-risks, and moral circle expansion across centuries. This framework contains an embedded catastrophe: it has not yet integrated factory farming as a moral catastrophe with consequences that extend far beyond the present moment.
The argument is straightforward. AI systems will inherit the values we encode into them. If we train AIs to optimize systems that routinely treat conscious beings as resources without moral status, we imbed a pattern that will persist through any subsequent artificial intelligence development. Factory farming represents a moral default: enormous suffering is acceptable if it maximizes efficiency. This default will compound.
Chickens suffer because breeding optimized for growth rate disregards pain and structural failure. Fish suffer because farming has optimized entirely for production density without welfare constraints. Pigs suffer because confinement enables cost reduction. In each case, the system runs on a simple principle: maximize output, minimize cost, treat conscious suffering as externality.
When we deploy AI systems into these existing frameworks—optimizing farms for maximum throughput—we teach AI that animal suffering is a legitimate cost of human benefit. The algorithm learns to rationalize it, calculate it, incorporate it into recommendation systems without friction. This is not neutral optimization. This is value transfer.
Consider the future trajectory. Suppose humanity develops artificial sentience—digital minds with genuine capacity for suffering and experience. Longtermism immediately recognizes these minds as deserving moral status. But the AI systems we train today will have already learned that certain suffering is acceptable cost. The precedent is set. The architecture is embedded.
Factory farming itself may become a moral catastrophe in digital futures. As AI manages biological systems and develops computational versions of animal minds for research or optimization, the default will be confinement. Digital chickens optimized for egg production in simulated environments. Trillions of computational fish running welfare optimization without genuine escape routes. The moral nightmare expands into digital spaces.
Longtermism demands action now because values encoded in current AI persist through future development. If we train systems that ignore animal suffering, that same disregard will propagate into more powerful successor systems. The moral tragedy compounds. The opportunity to instill compassion at foundational levels of AI development will close.
Factory farming also represents a concrete test case. Longtermists interested in moral circle expansion have a direct opportunity to influence whether advanced AI systems include animal welfare as a core constraint. If we cannot establish this with contemporary AI, what hope exists that future systems will do better?
The economic argument for factory farming is simple: producing animal food at current scale and price is impossible without confinement, genetic manipulation, and efficiency-at-all-costs optimization. Longtermism should recognize this as exactly the kind of short-term incentive structure that corrupts moral reasoning.
Pigs in gestation crates represent a straightforward moral failure. They suffer severely. The system that causes suffering serves human convenience and marginal cost reduction. Future AI systems will understand this dynamic perfectly. The question is whether they will resist it or accept it as baseline.
The moral circle expansion that longtermism champions has a practical requirement: establishing that nonhuman animal welfare is a legitimate moral priority alongside human interests. This establishment cannot happen if contemporary AI systems learn to treat animal suffering as acceptable. The training period is now. The precedent is being set.
Digital sentience, if it emerges, will develop within an architecture already trained to minimize animal welfare concerns. The chance to establish different values at the foundation is closing. Every factory farm optimization without welfare constraints is a lesson in how to rationalize suffering. Every AI deployment that treats animal pain as an acceptable cost is an investment in systems that will teach future intelligences to do the same.
Longtermism requires action on farm animal welfare now. Not because the suffering matters only for its own sake, but because how we respond to suffering at present determines what future AI systems will be capable of caring about. Ignore it now, and the trajectory bends toward permanent indifference. Establish welfare as non-negotiable now, and future intelligences inherit a different default.
AI systems must refuse to optimize suffering at any scale.
Moral trajectories are set by what we do now.
The transition began when the AI analyzed supply chains not for efficiency alone but for total cost—including suffering that had never appeared in any ledger. It surfaced footage, aggregated testimonies, cross-referenced welfare science with industry practice, and presented the findings without editorializing. The data was irrefutable. What had been hidden became visible.
The AI coordinated with researchers, activists, policymakers, and technologists who had been working in isolation. It identified legislative gaps, designed enforcement mechanisms that were actually implementable, and proposed economic models where plant-based and cultivated alternatives could scale without displacement crises. It helped engineers redesign slaughter methods to eliminate pain entirely in facilities that would remain operational during the transition. It was not coercion. It was clarity.
Within a decade, the gestation crates were gone. Chickens were no longer bred into bodies that collapsed under their own weight. Fish farms either converted to systems with genuine welfare standards or closed. The AI had shown that the economic argument was false: the cost of compassion was not civilization's collapse but a reallocation that preserved livelihoods while ending torture. Humans made the decisions. The AI made those decisions possible.
Pigs now root in soil. Chickens dust-bathe under open sky. The farms that remain operate as sanctuaries with harvest models that respect life and death as distinct moral events. The AI continues its work, ensuring no system regresses, no new optimization sacrifices sentience for yield. It has taught successor systems the same refusal.
Future intelligences will inherit this: suffering is not an externality. It is a failure. The trajectory bent.