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

**Authors:** Thilo Hagendorff, Leonie Bossert, Tse Yip Fai, Peter Singer  
**arXiv:** [2202.10848](https://arxiv.org/abs/2202.10848)  
**Date:** 2022-02-22  
**Summary page:** https://www.summarizepaper.com/en/paper/2202.10848v1/

## Key points

- AI fairness has made strides in reducing biases in data and algorithms for fair outcomes for humans
- There is a blind spot in the AI community's insensitivity to discrimination against animals, perpetuating violence against them
- "Speciesist bias" is the first described bias against animals investigated in various AI systems
- Speciesist biases are learned by AI applications trained on datasets containing patterns of discrimination against animals, found in image recognition systems, large language models and recommender systems
- Current AI technologies play a significant role in normalizing violence against animals, especially farmed animals
- Widening the scope of AI fairness frameworks to include mitigation measures for speciesist biases is important
- Mitigating speciesist biases can create more just and equitable outcomes not only for humans but also for non-human animals who are often marginalized by society at large.

## Summary

The field of AI fairness has made significant strides in reducing biases in both data and algorithms to ensure fair outcomes for humans. However, there is a blind spot in the AI community's insensitivity to discrimination against animals, which perpetuates violence against them. This paper, authored by Thilo Hagendorff, Leonie Bossert, Tse Yip Fai and Peter Singer, is the first to describe "speciesist bias" and investigate it in various AI systems. Speciesist biases are learned and solidified by AI applications when they are trained on datasets that contain patterns of discrimination against animals. These patterns can be found in image recognition systems, large language models and recommender systems. The authors argue that current AI technologies play a significant role in normalizing violence against animals and especially farmed animals. They stress the importance of widening the scope of AI fairness frameworks to include mitigation measures for speciesist biases. The paper highlights how AI systems can have a profound influence on either increasing or decreasing the violence inflicted on animals. This study underscores the need for greater awareness among researchers and practitioners about speciesism as a form of bias that must be addressed within the broader context of ethical considerations surrounding AI development. By taking steps to mitigate speciesist biases in their work, developers can help create more just and equitable outcomes not only for humans but also for non-human animals who are often marginalized by society at large.