Animal-Friendly AI NOW!
We stand at the precipice of a new era in artificial intelligence – and ethics has never been more important.
My investigations into ethical AI, in partnership with Sankalpa Ghose, Yip Fai Tse, Kasra Rasaee, and Jeff Sebo, have focused on an often-overlooked consideration: the impact of AI on animals. Our recent paper, "The Case for Animal-Friendly AI," proposes a new approach to AI development that incorporates animal ethics at its core by outlining how AI might plausibly represent the perspectives of animals. You can read our paper at www.animalfriendly.ai.
Our project (which is at an early stage, subject to ongoing research and validation) was driven by the realization that as AI technologies increasingly influence every aspect of our lives, the consequences on animal life are likely to be as – or even more – significant. Yet both ethicists and engineers have neglected this crucial aspect of AI development. We seek to change this and our work attempts to be a catalyst for the development of more advanced AI systems that consider the welfare of all sentient beings.
Our paper and our AI benchmark – which we call AnimaLLM – provide a framework and a software prototype for evaluating LLMs. iIt is an example of how philosophy can contribute to building a better future in which we extend our ethics and our engineering requirements beyond the narrow confines of speciesism and take seriously our responsibilities to all sentient beings. By demonstrating that it is both conceptually and technically plausible to assess AI technologies for their ability to extend ethics beyond members of the species Homo sapiens and accurately take the perspectives of animals into account, AnimaLLM serves as an exciting first step towards fostering a technology ecosystem that values the welfare of animals.
Our experiments with AI models like OpenAI ChatGPT and Anthropic Claude have revealed a wide variation in responsiveness towards animal welfare. These findings highlight current AI baselines and defaults regarding attitudes on animals, as well as identifying a method by which to investigate the capabilities and limitations of LLM prompts and responses involving non-human perspectives. Overall, our results, while preliminary, underscore the urgent need for AI developers, policymakers, and citizens, to prioritize Animal-friendly AI now.
As encouraging as our initial findings are, they illuminate the philosophical, technical and political challenges inherent in this endeavor. AI is a machine mirror of humanity, reflective of our best and worst attitudes as a society. I have been calling for Animal Liberation for decades now, and this work begins to ask the question of what that might look like in a Machine Age. The world is changing rapidly, and while the risks are often discussed, the opportunity to build a better future – which by any reasonable measure would mean less animal slavery – is one that gives me hope.
You can be part of the change. The next time you have a conversation about AI and Machines, make sure you ask: “What about Animals?”
Working for a more ethical future is not solely the responsibility of those directly involved in AI development. We want our paper to serve as a call to action: to reassess our relationship with technology and its impact on the natural world. By fostering greater awareness and understanding of the ethical dimensions of AI, we can create a more inclusive and compassionate future for all living creatures.
As we look to the future, the potential for AI to contribute positively to animal welfare is immense. From enhancing our understanding of animal behavior and needs, to developing technologies that can represent and protect animals across society and in the wild, AI has the potential to be a powerful ally in the struggle to expand the circle of protected beings beyond our own species. Making this happen will require that those who believe in ethics, reason, and respect for sentient life work together to ensure animal consideration is embedded as a top priority of AI development and use now.
The main thing is of course to make practical improvements and policy changes, but it is sometimes nice to engage in some utiopianism, and in that regard I highly recommend Becky Chamber's novella 'A Psalm for the Wild Built,' where one of the main characters is an embodied AI with a passion for wild animals :)