Are Vegans Getting AI Wrong?

Animal agriculture is using AI to maximise profits, many vegans are refusing to use it on principle. Someone's making a mistake.

While vegans debate whether AI is ethical to use, and many refuse to use it over environmental concerns, the animal agriculture industry has already built a $2 billion AI operation. I sat down with researcher and vegan Andy Masley to dig into the viral headlines about AI’s water use and environmental impact, and if refusing to use AI might be the biggest mistake the animal movement is making.

---

---

Andy Masley is an independent researcher and Substack writer who spent the last few years doing something most of us haven’t: fact checking anti-AI headlines. Andy is so confident in his research that he offers $300 to anyone who can find a significant error in his work. He’s already paid out once on an old post from when he was just starting out, and he stands by everything since.

This post covers the highlights of our conversation. We got into an interesting discussion near the end where one of his answers surprised me, it might surprise you too. I’ll leave that for the video, it deserves the full context and explanation from Andy himself.

“You’re showing up to a gunfight with a knife”

AI in animal farming is currently a $2 billion global market, and it’s growing at about 15% a year. That’s a conservative estimate. These industries are using AI to optimise feed, maximise output, and make animal agriculture more profitable. Meanwhile, some in the animal advocacy community won’t touch it.

Andy didn’t hold back:

“If they’re not using AI, I think at this point they’re at a disadvantage. And that would make me incredibly sad as someone who really wants our team to win.”

Andy has been vegan for ten years so he’s not coming at this from the outside. His point isn’t that AI is perfect or beyond criticism, it’s that refusing to use it while the other side scales up is, in his words, kneecapping ourselves.

The “AI uses a 500ml water bottle” claim - where did it come from?

This story was everywhere. Euronews, Forbes, the Washington Post. The claim made is that every ChatGPT conversation uses a 500ml bottle of water. It spread fast, appealing to the growing number of people feeling uncomfortable with AI.

Andy traced it back to its source, and the methodology doesn’t hold up.

“The article itself was assuming you’re sending like 10 to 20 prompts per email. And if you actually look into the methodology, there’s a lot of weird stuff going on. They seem to be assuming AI is running on hardware that was six years old at the time, and AI hardware becomes incredibly optimised, incredibly fast. Every five years or so, the average energy cost of doing the same thing on an AI chip drops by a factor of about a hundred.”

There’s more: they were also counting water used offsite, including water evaporated from lakes at hydroelectric plants, as part of the cost of the ChatGPT prompts.

“The (hydroelectric) plant would be there otherwise. The lake is also capturing a lot of water from rain. There are all these weird questions.”

Add it all up: calculations based on the wrong hardware, assuming people are sending ChatGPT 10-20 more prompts than needed to write an email, and attributing non-direct water use. When correcting for these errors, the real figure is closer to one millilitre per prompt. Not 500ml, one.

One hamburger = tens of thousands of ChatGPT prompts (in water)

This is a comparison I wanted Andy to talk about because I think it’s something the animal advocacy movement really should know.

“If you look at the cost of beef, especially where beef cows are raised on alfalfa, which is an especially water-hungry plant, the cost of any one beef burger is on the order of at minimum tens of thousands of chatbot prompts, even if you divide by the number of people eating the cow altogether.”

He was careful to flag he wasn’t engaging in whataboutism. He doesn’t say “you’re vegan so ignore everything else”, his point is more specific:

“Of all the people in the world, if vegans are feeling especially guilty about these incredibly tiny amounts of water that AI can add to your footprint, I think you’re kind of missing the bigger picture. It would take something like at least a thousand chatbot prompts per day to raise your daily water footprint by 1% because most of our water footprint isn’t actually in our homes. It’s mainly in the food we eat. The most efficient way to cut your water footprint is to be vegan. Those cuts matter a lot. Saying ‘never use AI’ is almost like saying ‘if you want to save money, buy one fewer gumball per month.’”

Golf courses vs. data centres

You’ve probably seen some scary headlines about AI data centres being built in deserts and other areas short on water. Andy made a comparison that gets almost no news coverage.

“Golf courses in America are currently using significantly more water than all the data centres in America combined, and generating way less tax revenue per unit of water. In Arizona, nearby golf courses use something like 30 times as much water as all data centres in the state.”

And then he brought it back to agriculture:

“Something like 20% of Arizona’s water is used to grow alfalfa, which is mostly used for animal agriculture, either in the US or shipped overseas. Almost all actual water issues are downstream of agriculture. It’s very easy to find other things that completely dwarf AI use almost wherever you go.”

Why doesn’t agriculture get the same scrutiny? Andy’s take was blunt: that farmers occupy an almost untouchable place in our culture. I agree, they’re almost viewed as gods to the average person. We don’t criticise the people who grow our food, and anyone who does faces the wrath of lobbyists and the public. This is something animal advocates understand better than most.

How to use AI as an advocate and where Andy draws the line

I asked Andy: what does he use AI for, and what lines won’t he cross?

On the lines he won’t cross:

“I don’t use it for any kind of social companionship. That specifically freaks me out quite a bit. I think the world in general is a very lonely place, and I worry that it really corrodes something fundamental to the social fabric.”

On the echo chamber trap:

“It’s very easy to make AI argue for anything you want. It’s still quite sycophantic. I think trying really hard not to use AI as a weapon that just hammers your opponents and instead asking it to challenge you is really important. Where might I be wrong here?”

That last point is something I feel strongly about for advocates. Before your next debate, online or in person, put your argument through an AI chatbot like ChatGPT, Claude or Google Gemini, and ask it to challenge you. Ask it where you’re weakest, ask it what the best counter-arguments are. This is something you literally couldn’t do before AI unless you had access to a seasoned advocate willing to spend at least an hour stress-testing your positions. Now you can do it anytime, any day, for free, and it takes a few minutes.

Andy put it well:

“There have been so many times when I’m writing something and I realise via AI that a central argument I’m making is incorrect. Every single time, it feels like a gut punch. You have to acclimate to that over time. At its best, AI can hopefully be a stress test of that impulse in ourselves, to resist just wanting to win any debate regardless of how correct we are.”

Is AI going to be net positive or negative for animals?

I saved this question for last in our conversation, and Andy gave an honest answer that I have a lot of respect for.

“I have absolutely no idea. The Industrial Revolution was on net just incredibly good for human lives, and I think it was just so clearly net bad for animals in terms of our new ability to farm animals at scale and factory farm them. If AI is just a continuation of that pattern, I can see so many worlds where factory farming just gets much more entrenched. It’s going to matter a lot what we actually do with it.”

And then this, which I think is a great takeaway from the whole conversation:

“I really need animal advocates thinking about this incredibly consequential new technology. If we’re seeding that ground and we’re all saying ‘oh, but it uses a few drops of water every time’, I think that’s a catastrophic mistake. Even if you use AI and come away hating it, it’s just so important to at least understand it. We really, really need you on the ball here.”

Watch the full conversation

In the video we explore these topics much further, getting into AI plagiarism, whether AI can genuinely be used to debate animal ethics, and the topic where Andy’s thoughts surprised me: who actually benefits from people being afraid of AI?

That last part gets a bit conspiratorial, I’ll let you make your own mind up. But I think you’ll enjoy hearing Andy’s take, and seeing where our opinions differ.

Andy’s Substack is at blog.andymasley.com, I recommend checking him out if you want all the data, and subscribe to stay informed. He plans to start writing on animal related topics too, which I’m looking forward to.

If you found this post useful, share it with an advocate you think should see it. I understand the concerns around AI, I share some of them too. But as animal advocates, we’re already at a massive disadvantage. We’re fighting for the most oppressed, most disregarded, exploited, abused beings in the world. I don’t think we can afford to be leaving such powerful tools on the table, especially when the other side is already using them against animals.