SCRIPT

These are times of change.
Artificial Intelligence is transforming the world as we knew it.
Human beings will not be the only ones affected.

Matti Wilks · University of Edinburgh: I think that in the same way that AI is going to affect so much of human life, possibly all aspects of human life, we can really think about it having similar impacts on animal lives.

Kevin Xia · Hive: What is upon us may be something as big and massively changing as the industrial revolution or the agricultural revolution.

Peter Singer · Princeton University: I think there’s a real concern that the effect of AI on animals will be negative.

Constance Li · Sentient Futures: I think it would be crazy not to wake up and see that this is actually going to be transformative. And the sooner we can do that, the better the world will go for animals.

Jonathan Birch · London School of Economics: We should be really really cautious and make sure that AI is being used in an ethically responsible way.

Peter Singer: AI is already having an impact on animals in various ways. Large language models are answering questions about animals, and sometimes they’re answering them in ways that make people think of animals as just objects to eat.
But the largest and most worrying impact that AI is starting to have is on factory farms, because there we’re talking about millions, tens of millions, hundreds of millions, and even billions of animals that are being affected, and more will be affected in future.
And the whole point of this will be to make factory farming more efficient. And that in itself is a bad thing, because in the efficiency of factory farming, the producers are not interested in the cost to animals.

Max Taylor · Animal Charity Evaluators: Think who’s going to be controlling these AI systems, and what are their motivations going to be, and we can look to the past to see what the motivations of the big producers are, where the vast majority of farmed animals are kept, and that tends to be just finance and productivity. So, companies are going to use it to increase their profits, to cut down costs, and get as much meat or milk or eggs from their animals as possible.

Constance Li: By default, we cannot assume that AI will go well for animals. If we look at history, every new technology that’s come by, like the Industrial Revolution, like the Green Revolution, where we planted more crops, all these things enabled factory farming.
And that went really bad for animals. And so right now, we’re seeing AI play out in industrial animal agriculture. They are using it more and more to automate these systems.
If we miss this opportunity, and this technology goes everywhere, it’s going to be incredibly hard to change.

Peter Singer: If AI makes factory farms more efficient, makes the products cheaper, it makes it more difficult for the plant-based alternatives to compete with them. And it also may mean that they can stock animals even more densely than they are now, because the AI maybe will alert them earlier to problems caused by the stocking density, so they’ll have a more precise measure of just how many animals you can crowd into these buildings.

Jonathan Birch: I think one of the big risks here is that we might see AI being used to drive up stocking densities. So we might be seeing more salmon packed into these pens because the technology makes that possible. We might see stocking densities that wouldn’t otherwise be viable.
Of course that has to be avoided. We might still be enabling the industry to grow, to increase in size. The number of salmon being farmed in these pens out at sea might increase and if we think that on the whole the lives of these animals are pretty bad because they’re living in really cramped conditions then the worry here is that we might be having more and more salmon living those really bad lives.
Those lives that are not worth living as a result of how AI has been used. So there’s a lot we need to be careful about here. We shouldn’t be hyping up this industry. We shouldn’t be boosterish about it.

Bernice Bovenkerk · University of Wageningen: There’s already big companies that are making agriculture facilities that are completely run by AI completely independently of humans. So AI checks the water quality, does the feeding? Does the slaughter, everything, and they’re building fully or semi submerged aquaculture facilities that can hold up to a million fish at a time.
So you could imagine, what if something malfunctions? If the AI system malfunctions, then a lot of fish will die at the same time.

Fai Yip Tse · National University of Singapore: AI can do a lot of stuff to increase the efficiency aquaculture of offshore aquaculture, and it might have a future of AI enabling a lot of offshore farms just covering a lot well, let’s say 1% of our ocean surface. That’s totally feasible. And actually that’s kind of efficient. Some of the advocates of the of offshore aquaculture is having, they want to have a lot more offshore fish farms out there. And one very big barrier they’re having now is labor. Because you, if you don’t run it with AI, you need people to live on that farm, but with AI, you can just automate it entirely without expensive labour. So I worry about that a lot.

Aditya S. Karanam · Animal Ethics: The current state of scientific evidence is that many invertebrates have the capacity to feel pain. And these are the groups of animals that are found in the highest number in factory farms today. In addition to this, AI driven genetic selection could lead to farmers optimizing for traits like rapid growth or higher protein content. Again, optimizing for these things could mean that the individual animals could end up suffering more throughout the farming process.

Jeff Sebo · New York University: AI is going to be increasingly important because AI will, for better or worse, be used to surveil and control and hopefully communicate with and better understand wild animals.
And we want to think about how we can secure those benefits while mitigating the risks for them.

Aditya S. Karanam: AI can boost our efforts to better understand and also prevent wild animal suffering, both by anthropogenic harms as well as natural harms, for example, diseases, parasitism, hunger, thirst, accidents, and natural disasters. There’s a common objection that’s given against helping animals in the wild, which is that we will never have enough knowledge to go out there and help animals there in the wild. However, this objection is very pessimistic and over-complacent. At any rate, AI can make a significant difference in our efforts to help animals in the wild. It could be either through boosting the research to better understand the situation of animals and how we can help them, or through monitoring.

Oscar Horta · University of Santiago de Compostela | Animal Ethics: In addition to allowing us to measure the wellbeing of wild animals at the population level, AI systems can also make it possible to monitor the situation in which they are and also the impact of the different initiatives and measures to help them. This is crucially important because it is not just that they often are in really, really harsh situations; it is also that the vast majority of sentient beings on Earth are wild animals.

Jeff Sebo: There will also be unintended effects.
For example, self-driving cars might be trained to avoid human pedestrians more than non-human pedestrians. And so they might end up killing a lot of wild animals unless they are trained to avoid them as well. And there’ll be indirect effects.
For example, we know from the human case that algorithmic bias can amplify biases that already exist in the training data, racism, sexism, ageism, ableism. We can expect that that will be true for speciesism as well, that AI systems will amplify biases that we have against non-human animals, again, unless they are specifically trained to avoid those biases.

Oscar Horta: Speciesism is the discrimination against those who aren’t classified as members of a certain species. And human beings are speciesist all the time against other animals: they harm them in terrible ways; they refuse to help them even when they could do so very easily. And here the problem is that these speciesist attitudes could be inherited, so to speak, by AI systems. And this is something that we are seeing right now with frontier AI systems, which are LLMs. So that’s why it is very, very important to change this and we have a window of opportunity for this that may be closing soon.

Yip Fai Tse: One thing I worry is that large language models might have more and more power in shaping people’s value, including kids, because LLMs are also used in education.
And we can see a lot of species’ biases against animals, bad attitudes or bad impression regarding animals. So for example, if you ask ChatGPT or Claude for a random recipe, it’s most likely to be non-vegan. And also if you ask vague questions such as, what is good about chicken? And they might say things like chicken is good for food purposes, like as if the word chicken refers to just food stuff.

Vanessa Sarre · Macroscopic Ventures: We’re already seeing LLMs and the tools that LLMs allow us to create, we’re already seeing them being integrated into all parts of people’s lives
So that means the assumptions that are built into these models are really important. So for instance, if we have an AI model that treats animals as having moral weight and gives them importance then that can change. I believe that will change how an LLM will respond to questions about animals, how it will communicate with the users about animals and therefore how people view animals and the moral consideration that they give them.
These sorts of things are likely to have an impact on how the population thinks about animals and so we really need to be building the values that help animals and help them be seen as beings with moral weight. We really need to be building those values into LLMs from the get-go.

Adrià Moret · University of Barcelona: Many more animals may exist in the future, both in the mid-term future and the long-term future than those that exist right now. And there are some kinds of actions that we can presumably take that may influence the welfare and the interests of all these animals. And so doing these kinds of interventions and thinking more in the future, not just right now, could have very large impacts on the welfare of animals in total. And more specifically, it might make sense to have a more forward-looking approach in animal advocacy, trying to prevent certain harms from occurring instead of only a reactive approach that is about trying to reduce currently existing harms.

Adrià Moret: Right now, various AI companies are developing powerful AI systems, and their goal is that these AI systems are developed in a way that have very significant impacts in society in general. And we can expect the values of these AI systems to influence societal values in general, given that humans may be interacting with them constantly.
These AI companies are deciding right now, they are making ethical decisions about what kinds of values they implement into these AI systems.
They have certain documents that they use to determine these values with certain principles and phrases. And so we may have a very good opportunity right now to try to include some greater concern for animals there, an opportunity that we haven’t had in the past and that we may not have in the future, given that the values that we implement right now might be very difficult to change in the future, as many people have argued, given that, for example, AI systems might try to preserve their values in some ways, or that in the future, there might be much more interest in trying to determine the values that these AI systems have by many different actors. And this could make it so that in the very near future, maybe in the next five years, it could still be much, much more difficult to influence these AI companies, for example.

Caatje Kluskens · University of Wageningen: Language is often used to distinguish humans from animals by showing that they can communicate with each other. We can show that there are individuals that have needs, wants and desires. Animal communication research projects like the SETI project and the Earth Species project are developing AI to decode animal communication.
So before, animal communication researchers had to manually label all the data and with the use of AI this can be done automatically on a large scale.
The AI can maybe find patterns that humans weren’t capable of seeing in the large data sets.
We potentially could raise some empathy in the general public to basically show the general public that animals are also just like us in a way.

Sam Tucker · Open Paws: One of ways we often justify our speciesism as humans is saying well we’re more intelligent than animals therefore we can do what we want with them. Now imagine a world where we train that same way of thinking into an AI system. How do we think the AI is going to treat us? Probably not very well.
But if we teach instead that intelligence doesn’t matter, that sentience is what determines moral worth and that all living creatures deserve that basic respect and dignity, then that’s the values we’re going to embed into these very intelligent AI systems.

Constance Li: There’s also the possibility that digital sentience might emerge. So as these AI systems get more and more complex, at what point would the lights start to turn on? And they could become moral patients.
And humans would have to then grapple with, oh, here’s this non-human being that is more powerful and more intelligent than me. So I can’t use the usual justifications for exploiting them. How do I then treat them? Should I expand my moral circle, or should I plant my feet on the ground and still say: “It’s only humans that matter?” And if they have to expand their circle to include digital minds, then maybe along the way, they could consider that animals deserve moral consideration as well.
So I think AI can really expand your mind outside of the box.

Bernice Bovenkerk: AI could also be used by animal advocacy organizations, for example, to help them to run smoother campaigns, or to help them to work together better, or it could help in creating alternatives to animal products.
And so there’s a lot going on in terms of creating cell-based meat, or in vitro meat, or whatever you want to call it, and AI might also be employed for that to help us, well, develop that product much quicker, or to upscale it, for example.
Yeah, so those are just a couple of things that I’m thinking about. But of course, there are many different ways in which we can think about how AI might actually best the lives of animals.

Kevin Xia: AI could help us in being generally more productive in our work. We could use AI to massively individualize mass outreach campaigns. We could use AI to reduce the cost of alternative proteins.
Or we could even use AI to research effective ways to do animal advocacy. However, a lot of these interventions can be very easily mirrored by the industry themselves. The industry too can and likely will use AI to make their companies more productive, to massively individualize marketing campaigns, to reduce the cost of animal products, and to research ways to defend against effective ways of doing animal advocacy.
I believe it may turn out to be very important for us to run all of these interventions still, just in order to not fall behind much, but to also leverage our comparative advantages.

Matti Wilks: If you have worked for a cause and something has functioned for you in a certain way for a really long time, it can just be really difficult to adapt to these new kinds of ideas and new kinds of methods. We would probably see this more generally. But I think also when you’re working with something that’s very important to you and very moralized, it can feel quite threatening for something like that to be changed. And given just the scope of how impactful AI could be and the potential harms that could come with it, as well as the potential benefits, I imagine it’s something that’s just really challenging for people to engage with.

Sam Tucker: Imagine what we could do with a hundred, a thousand times more people hours working on things. As we increasingly develop agentic AI that can take action on a user’s behalf, we’re going to exponentially increase our capacity. So the question that I think we’re all going to have to ask ourselves as animal advocates is what would it look like if I could manage a team of a hundred, a thousand different advocates to work on different things for me? And that’s where AI agents are going to play a huge role.
You can imagine in the early days of the internet people thought this is just a fad, this is something for nerds, it’s not going to catch on. But the people who had that kind of attitude and didn’t adapt to the new technology were obviously going to be a lot less effective than the people who did. So imagine if we could go back in time to the early days of the internet and be the first to be doing things like search engine optimization, to get animal advocacy issues, ranking and search engines.
There’s so many possibilities of what we could have done to be more effective if we knew this technology was going to be so massive. We’re at a point now of AI where we do know the technology is going to be huge in the effects on society and really we have to start taking action now rather than waiting until it’s too late.

Constance Li: AI is going to do a lot of the ground work for us in the near future. And the people that can wield it the best, that have a vision and can manage a bunch of little AI agents to make it happen, those are the people that are going to have a lot of control over the future.
So the sooner that we can take advantage of AI for animals, the better the future might go for animals.

Max Taylor: I think the number one thing is just really getting people comfortable with the idea of experimenting with AI as much as possible. That’s probably like one of the number one things that I’d recommend, especially as things get more and more advanced. Because, you know, I’m finding that my research output has doubled just with current AI tools, but they really are increasing exponentially in terms of how useful they are.
So in a year’s time, maybe it’ll be kind of quadruple or quintuple the output. So I think keeping on top of those sorts of things is what really is going to move research organisations forward, or probably all animal advocacy organisations forward. So it’s something to start really getting into grips with now.

Alexandra Hammond · Liverpool John Moores University: At the moment with the advancements in AI, I think it’s a prime opportunity to be able to actually take concrete steps to be able to replace animals in testing. Unfortunately it is a fairly under-researched area at the moment. So some of the ways that we may look at replacing the animals in testing is by using AI to develop a model that can predict how toxic substances are.
So that’s something that I look at at the moment. I would really like to be able to use those models for cases where animals experience the most severe suffering and the experiments that cause the most prolonged suffering.