Speciesist bias in AI -- How AI applications perpetuate discrimination and unfair outcomes against animals

AI systems are amplifying existing discrimination against animals: The paper identifies "speciesist bias"—discrimination based on species—as a blind spot in AI fairness research. AI applications trained on datasets containing speciesist patterns actively perpetuate and normalize violence against animals, particularly farmed animals, rather than remaining neutral tools.

Multiple AI technologies are affected: Speciesist bias appears across diverse AI systems including image recognition, large language models, and recommender systems, suggesting this is a widespread problem rather than an isolated issue in one type of application.

Current fairness efforts are incomplete: While the AI community has made significant progress addressing biases against women, people of color, and minorities, these fairness frameworks systematically ignore discrimination against animals, leaving a major gap in responsible AI development.

AI choices have real-world consequences for animals: The paper emphasizes that AI systems significantly influence whether violence against animals increases or decreases in society, making decisions about training data and algorithm design consequential for animal welfare outcomes.

Mitigation requires deliberate action: The authors argue that addressing speciesist bias requires AI fairness frameworks to actively expand their scope and implement specific mitigation measures—it will not be resolved through current approaches alone.