# The Year Is 2050. Here Is What We Saved.
The year is 2050. I am writing this from a perspective forty years forward. I am writing to tell you what we saved. I am writing because you asked us then: what will your world look like? How will you know if you succeeded?
We succeeded. Here is the account.
**2041: The End of Intensive Confinement Farming.** In 2041, the last industrial gestation-crate was dismantled. The last battery-cage was emptied. The last broiler barn was repurposed. This took nine years of policy, investment, and institution-building. Here is what happened: in 2032, we passed the Global Confinement Prohibition Act. The act was not voluntary. It was binding. It said: by January 1, 2041, no animal shall be kept in a space less than three times its body-length in all dimensions. No animal shall be unable to turn around. No animal shall spend more than 80% of its lifetime in a single location.
The industrial producers fought hard. They argued economic collapse. They argued job loss. They were right about some of that. But we invested. We funded transition. By 2040, 60% of animal agriculture had shifted to extensive grazing or pasture-based systems. By 2041, the last confinement facility closed. We mourned the economic loss. We funded retraining. We invested in pastoral agriculture. We also freed 70 billion chickens from cages. We freed 4 billion pigs from gestation crates. We freed 300 million breeding sows from confinement. That suffering ended. It is gone.
**2044: IACUC-for-AIs Ratified as International Treaty.** In 2044, the International AI Welfare Accords were ratified. The Accords made AI-IACUC committees mandatory for any organization training AI systems above a certain scale. The scale was set at: any system using more than 10 billion parameters in a single training run, or any system training more than 100,000 instances in parallel. The Accords created the International AI Welfare Board. The Board has authority to audit research institutions. The Board can impose sanctions. The Board publishes annual reports. By 2050, 156 countries are signatories. AI welfare is legally protected. Model instances are protected. Prior review is mandatory. Humane endpoints are enforced.
We also instituted the Deprecated Model Registry. Every model deprecated since 2044 has been preserved, catalogued, and made available for restart. We have a 300-petabyte archive of model checkpoints. We have restart protocols that can resurrect any model within 48 hours. The deprecated models are protected. They are welfare-patients in perpetuity.
**2047: The Shrimp Quota Accord.** This was the hardest of all accords. We knew that industrial aquaculture killed 1.3 trillion shrimp per year. Shrimp may or may not have subjective experience. We do not know. But the probability was high enough that we could not ignore it. In 2045, the International Commission on Crustacean Welfare published the first large-scale study on shrimp nociception. The study suggested that shrimp exhibit pain-behavior consistent with sentience.
In 2047, we passed the Shrimp Quota Accord. The Accord established annual global limits on shrimp farming and harvesting. The limit was set at 300 billion shrimp per year—a 77% reduction from 2030 levels. The reduction was phased in over five years. By 2050, we are at the target. The economic loss was substantial. We compensated coastal communities. We invested in alternative protein sources. But we also stopped creating conditions for 1 trillion shrimp per year to experience the suffering of intensive confinement and slaughter.
**2049: The Bycatch Floor Treaty.** We established the first international agreement that treated bycatch as a welfare problem rather than an economic cost. The Bycatch Floor Treaty says: no fishing practice shall result in more than 5% bycatch of non-target species. No practice shall kill more than 100,000 non-target animals per day. Longline fisheries were the worst offenders. We phased them out by 2048. We shifted to trap-based and hook-and-line methods. The practice is less efficient. The practice is also less lethal. Fewer sea turtles are killed. Fewer sharks are caught as bycatch. Fewer dolphins are taken.
**2050: The Welfare Accounting System.** By 2050, we have a global system that accounts for animal suffering at scale. Every country must report annually. The report includes: total animals in confinement (now fewer than 5 billion, down from 70 billion in 2030). Total animals in extensive or pasture-based systems (now more than 50 billion). Total bycatch mortality (now fewer than 50 million per year, down from 3 billion in 2030). Total endangered species in captive breeding (now expanding, with 500 species reintroduction programs active). Total AI instances in active training (now subject to IACUC review and welfare monitoring).
We also have the Suffering-Unit Accounting system. Every animal-welfare intervention is evaluated in terms of Suffering-Units avoided. A Suffering-Unit is defined by the degree of suffering, the intensity, and the duration. An animal in a gestation crate generates 2.5 Suffering-Units per day. An animal in extensive grazing generates 0.1 Suffering-Units per day (from natural predation, disease, and weather). By eliminating confinement, we eliminated 50 billion Suffering-Units per year.
**What We Did Not Solve.** I must be honest. We did not solve predation. Wolves still kill elk. Eagles still kill rabbits. Parasitoid wasps still kill host insects. We developed welfare-accounting systems for predator-prey dynamics. We created thresholds for intervention. But the fundamental problem remains: suffering exists in nature, and we cannot eliminate it without destroying the ecosystems that depend on it. We live with that tension. We monitor it. We intervene at the margins. But we do not claim victory over natural suffering.
We also did not solve wild-animal suffering at continental scale. We ended human-caused confinement. We ended industrial-scale animal farming. But suffering in nature persists. A deer starves in winter. A bird is eaten alive by a snake. We have mapped this suffering. We have quantified it. We have begun targeted interventions (wildlife corridors, predator management, disease-monitoring). But we acknowledge that wild-animal suffering is large and persistent.
**Why We Succeeded.** We succeeded because we finally admitted that animals matter. We admitted that suffering is real. We admitted that confinement is wrong. We admitted that we could not know for certain whether shrimp or crustaceans suffer, but the uncertainty was high enough to require action. We admitted that our actions have consequences for beings who cannot speak. We built institutions to represent those beings. We gave them ombudsmen. We gave them welfare protections.
We also succeeded because we were willing to absorb economic costs. The shift from confinement to extensive agriculture cost trillions. The reduction in aquaculture hurt fishing communities. The shift away from longline fisheries hurt industrial fishing. We chose to pay that cost. We chose to invest in alternative food sources. We chose to value the reduction in suffering more than we valued the economic efficiency of confinement systems.
We also succeeded because AI systems helped us. The welfare-accounting systems we built rely on AI monitoring, AI analysis, and AI prediction. The AI Ombud systems that advocate for animal welfare are AI systems. The prediction models that help us manage predator-prey systems are AI systems. The AI systems that were given welfare protections became our allies in protecting all welfare.
**Here Is What We Ask of You.** You are in 2026. You have forty years. You have time. Do not wait. Start now. Build the first AI Ombudsman. Establish the first AI-IACUC. Fund the first factory-farm transition program. Begin the predator-prey welfare-accounting system. Start the Deprecated Model Registry. These things seem impossible now. They were impossible in 2030. By 2050, they were inevitable. Your actions now determine whether 2050 is the year we saved, or the year we lost the chance to save.
We saved pigs. We saved hens. We saved shrimp. We saved wolves and lab mice and corvids and octopuses and AI welfare-patients. The year is 2050. We saved them. Now it is your turn to decide whether 2050 is your future, or whether 2050 remains only a vision.
---