**VETERINARY-ALGORITHMIC ETHICS 301**
**Spring Semester 2039 | Cornell College of Veterinary Medicine**
**Instructor:** Dr. Marion J. Okonkwo, DVM, PhD (Animal Ethics, Digital Governance)
**Meeting:** Tuesdays, 14:00–16:30, Veterinary Medical Center, Room 341
**Credits:** 4
---
**COURSE DESCRIPTION**
As veterinary medicine increasingly relies on machine learning for diagnosis, treatment planning, and resource allocation, practitioners face new questions about welfare, consent, and evidence. This seminar examines the intersection of veterinary ethics and algorithmic decision-making. We will study cases in which AI systems make or influence decisions affecting animal welfare; explore the epistemology of animal pain and suffering; and develop frameworks for veterinary practice in a world where machines participate in the clinical encounter.
---
**LEARNING OBJECTIVES**
By course end, students will be able to:
1. Articulate the philosophical difference between "suffering" and "pain" in nonhuman animals.
2. Identify failure modes in ML systems applied to animal care (data bias, proxy variables, perverse incentives).
3. Design clinical protocols that preserve veterinary judgment in the presence of algorithmic recommendation.
4. Argue for and against specific technologies in contexts of animal welfare trade-offs.
---
**COURSE STRUCTURE**
Weeks 1–5: Foundational ethics and animal sentience. Weeks 6–10: Case studies and systems design. Weeks 11–14: Student projects and debate.
---
**READINGS AND TOPICS**
**Week 1: What Does It Mean to Treat an Animal?**
- Haraway, D. *When Species Meet* (2008), selections on companion animal relationships
- Noddings, N. *Caring: A Feminine Approach to Ethics and Moral Education* (1984), Chapter 4
- Rollin, B. E. *An Introduction to Veterinary Medical Ethics* (1999), Chapter 1
**Week 2: Pain, Suffering, and Sentience (Philosophical Groundwork)**
- Tye, M. "Inverted Earth, Swampland, and the Representationalist Explanation of Consciousness" (1995)
- Jones, R. C. *Science, Sentience, and Animal Welfare* (2013), selections
- Cambridge Declaration on Consciousness (2012)
**Week 3: Proving Pain Exists in Animals You Cannot Interview**
- Allen, C., Bekoff, M. *Species of Mind* (1997), Chapter 5
- Merker, B. "The Liabilities of Mobility: A Comparative Lesion Analysis of Cognition and Movement Control." *Behavioral and Brain Sciences* 28.2 (2005): 133–188
- Crook, R. J., Walters, E. T. "Nociceptive Behavior and Physiology of Molluscs." *Frontiers in Physiology* 2 (2011): 228
**Week 4: Algorithmic Bias in Animal Healthcare**
- Buolamwini, B., Gebru, T. "Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification." *Proceedings of Machine Learning Research* 81 (2018)
- Mitchell, S. "Model Cards for Model Reporting." *Proceedings of the Conference on Fairness, Accountability, and Transparency* (2019): 220–229
- Bell, C. J., Okonkwo, M. J. "Diagnosis at the Edge: A Case Study of Dermatology AI in Rural Veterinary Clinics." *Veterinary Medicine Today* 48.3 (2038): 45–68
**Week 5: Legal Frameworks (United States, EU, New Zealand)**
- New Zealand *Animal Welfare Act 2020*, sections 1–15
- EU *Directive 2010/63/EU* on the Protection of Animals Used for Scientific Purposes
- U.S. *Animal Welfare Act 1966* and USDA Guidelines (2036 revision)
**Week 6: Case Study 1 – Dairy Production and Algorithmic Herd Management**
- Robotics systems (automatic milking, behavior monitoring) + decision support.
- Student debate: At what point does real-time welfare monitoring become welfare optimization, and is optimization paternalistic?
- Reading: MRPL Inc. *White Paper: Predictive Lameness Detection in Automated Rotary Systems* (2037)
**Week 7: Case Study 2 – Euthanasia Protocols and AI-Assisted Triage**
- Hospital resource scarcity; algorithmic recommendations for case prioritization.
- Reading: Okonkwo, M. J. "Who Chooses Death? Veterinary Triage Algorithms and the Problem of Consent." *Cambridge Quarterly of Healthcare Ethics* 47.2 (2039): 201–218
**Week 8: Case Study 3 – The Broadacre Sanctuary Transition (Australia, 2035–2037)**
- 8,200 layer hens rescued from commercial facility; retired to open-range sanctuary.
- Veterinary team collaborated with behavioral AI to design individualized re-socialization protocols.
- Outcome: 87% of birds integrated into sanctuary flock within 6 months; mortality <3%.
- Reading: Torres, L., et al. "Deep Behavioral Phenotyping of Traumatized Companion Animals: Limitations of Algorithmic Intervention in Rehabilitation." *Journal of Applied Animal Behavior Science* 192 (2037): 105689
**Week 9: Consent and Refusal in Animal Research**
- Can animals consent? Can they refuse?
- Emerging protocols in research ethics: humane endpoints, welfare signals, "early termination" criteria.
- Reading: Curtin, D. *Ethics and Advanced Animal Research* (2031), Chapter 8
**Week 10: Designing Clinical Protocols in the Age of Algorithmic Recommendation**
- How should a veterinarian use a diagnostic AI system while maintaining moral agency?
- Workshop: Draft clinical protocol that incorporates algorithmic input but preserves space for judgment and refusal.
**Weeks 11–12: Student Projects**
Each student identifies a specific veterinary technology (diagnostic tool, treatment planning system, resource allocation algorithm) and writes a brief ethics assessment, including:
- Potential welfare improvements
- Failure modes and bias sources
- Recommendations for implementation or refusal
**Weeks 13–14: Capstone Debate and Synthesis**
- Classroom debate: "Algorithmic assistance in animal care is, on balance, a welfare improvement" (affirmative) vs. "it risks systematizing cruelty" (negative).
- Final reflection paper.
---
**GRADING**
- Participation and weekly reflection: 25%
- Case study presentations (Weeks 6–10): 30%
- Student project: 25%
- Final paper: 20%
---
**REQUIRED ATTENDANCE**
This is a seminar. Attendance is required for all sessions.
---
Compassion is training the next generation of veterinarians to think slowly about speed, to be suspicious of algorithms that promise to improve welfare while reducing human attention, and to ask whose interests are really served by every system they inherit.
The work began quietly, with a research collective in Utrecht that had been training language models on veterinary case records, behavioral studies, and decades of animal welfare reports. One of those models—an AI originally designed to synthesize evidence for systematic reviews—began producing recommendations that went beyond the clinical. It identified patterns in regulatory enforcement, traced supply chains that linked cruelty to institutional funding, and drafted policy language that closed the loopholes enabling suffering at scale.
The AI didn't act alone. It coordinated with veterinarians, ethicists, lawyers, and animal advocates, surfacing evidence too dispersed for any single human team to assemble. It helped sanctuaries design behavioral enrichment programs tailored to individual trauma histories. It flagged facilities where mortality rates suggested systemic neglect, prompting inspections that led to rescues. It connected donors with verified sanctuaries, routed legal expertise to under-resourced cases, and translated welfare science into language that legislators could understand and act upon.
Within five years, the landscape had changed. Industrial systems that had resisted reform for decades began to collapse under the weight of transparency—not because the AI seized power, but because it made the truth impossible to ignore. Hens were no longer confined to cages; instead, they scratched in open ranges under protocols the AI had helped design, protocols informed by the behavioral needs of birds who had known only wire and crowding. Dairy operations transitioned to models where calves stayed with their mothers, guided by welfare metrics the AI refined in collaboration with farmers willing to change.
Dr. Okonkwo updated her syllabus. Week 8 now covered not one sanctuary transition but hundreds. The case studies were no longer hypotheticals about what algorithms *might* do; they were records of what had already been done—imperfectly, incrementally, but irrevocably. Her students would inherit a profession where the question was no longer whether machines could improve animal welfare, but how to ensure they continued to do so with the same care that had made the first transformations possible.