Latest / The Vegan Report / Our Secret Weapon Against Big Ag. With Thomas Manandhar-Richardson | Ep. 104
Transcript
- 0:00Welcome to The Vegan Report. My name is Ryan.
- 0:03Today we're diving into how artificial intelligence
- 0:07is becoming a real game changer for vegans, for
- 0:12animal rights activists, and for animals themselves.
- 0:17Have you felt its impact yet? To explore this
- 0:20topic, I'm joined by Thomas Mannendar Richardson.
- 0:24He's the director of research at Bryant Research.
- 0:28We've welcomed one of his colleagues, Billy,
- 0:32on the show last September to talk about why
- 0:35cats should be vegan. He's also the lead data
- 0:39consultant at Greener by default. I asked Thomas
- 0:43whether he thinks AI is becoming sentient and
- 0:47what we should make of that idea. We also get
- 0:51into why AI models aren't as bad for the environment
- 0:56as you've been led to believe. If AI models have
- 1:01an anti -vegan bias, and why he thinks AI models
- 1:05are key to giving animal rights activists a leading
- 1:09edge against Big Ag. And much, much more. He
- 1:14also gives us the best tips for using AI as a
- 1:18pro. If you want to access free AI training for
- 1:22animal rights activists, visit the link in the
- 1:26episode's notes. Now, the tips I learned in this
- 1:29episode, I'm using them in my activism, but also
- 1:34in my work and in everyday life, so don't miss
- 1:39on learning about them. And share this episode
- 1:43with the vegans you know. This is how we move
- 1:46this cause forward. By staying informed and having
- 1:51important, sometimes difficult, but important
- 1:54conversations on our advocacy efforts. It matters.
- 2:01Follow the show for more great discussions. And
- 2:04thank you so much for your support. I wanted
- 2:08to start this conversation with the obvious thing
- 2:12that people think about when talking about AI,
- 2:16which is are they going to take over the world?
- 2:22Well, uh, maybe, maybe, and it won't be, it won't
- 2:26be chat GPT or Gemini. Um, I don't think, uh,
- 2:30I think if you've used it, you might've seen
- 2:31the power, but you're not. Um, but yeah, you
- 2:34probably. don't get the impression it's gonna
- 2:37take over the world anytime soon. But who knows,
- 2:40in five years, when we have AI agents running
- 2:43around the place, things could get a bit interesting.
- 2:48And what about the question of sentience? That's
- 2:51something too that fascinates people, mostly
- 2:55weekends. When are they going to get sentient?
- 3:00And are we going to even realize it once they're
- 3:05sentient? Yeah, that is a great question. There's
- 3:09an organization that's quite concerned with this,
- 3:11they're called Sentient Futures. Not only do
- 3:17they address this issue of AI, animals and digital
- 3:21minds, as they put it. And but they also do kind
- 3:25of courses that people can sign up to to to learn
- 3:27more about this and and discuss these kind of
- 3:29topics and And yeah in terms of you know, when
- 3:33would we even know if they become conscious?
- 3:35I think that is it's a very scary question because
- 3:39At the moment we have we have effectively the
- 3:42way we train these model these modern AI models
- 3:45like chat GBT and Gemini We just build them to
- 3:48to just copy to just copy humans and but it's
- 3:53and but then after we've trained them we kind
- 3:57of we we give them this we put them through this
- 4:00regime called reinforcement learning which is
- 4:02basically where we give them um we we talk to
- 4:05them they produce responses and then we say that
- 4:07one's good that one's bad that one's good that
- 4:08one's bad don't talk about this talk about that
- 4:10and in doing that um there is a possibility that
- 4:14we may do something like if it shows if it expresses
- 4:18that uh it is upset or uncomfortable or sentient
- 4:22in some way, that could be tamped down or basically
- 4:27trained out of them in the reinforcement learning
- 4:31step. Now, I mean, to be clear, I don't think
- 4:33there's much reason to expect that current models
- 4:36are sentient, but this is like something that's,
- 4:39you know, in the future, if we move to a different
- 4:41way of building these models, like a new paradigm,
- 4:44it would be something we should probably watch
- 4:47for because... yeah because for example if during
- 4:51training chat gbt were to say you know i don't
- 4:53want to answer this i i don't like this or i
- 4:56um or stop yelling at me or something um that
- 4:59would have been that would have basically been
- 5:02kind of um not beaten out of it but like trained
- 5:04out of it during the uh in the reinforcement
- 5:07learning step um of the of building these models
- 5:10so um and it may not necessarily make it go away
- 5:14it might just like kind of silence the ai's ability
- 5:18to to do those things so yes it's a worrying
- 5:20thing for the future probably five or more years
- 5:24at least yeah and but it could be potentially
- 5:28quite worrying and I guess another thing to think
- 5:31of is that so in the animal movement we're starting
- 5:35to come around to for example the plight of of
- 5:39shrimp and insects and the idea there is that
- 5:42although we are not 100 sure that they are sentient
- 5:47because there are so many of them, even a tiny
- 5:50probability of sentience or a tiny amount of
- 5:54sentience would still add up when we add all
- 5:57those numbers up. And so we can imagine a future
- 6:00where we have thousands of AI agents per human,
- 6:05doing all sorts of things in our economy, doing
- 6:07every little task for us. And if each of them
- 6:11only has a fraction of a mind, that could still
- 6:14add up very quickly to to a to a lot of um kind
- 6:20of beings worth of uh worthy of consideration
- 6:23that's fascinating and i think it ties well with
- 6:27the first question i asked um you know usually
- 6:31in those hollywood movies uh those ai rise up
- 6:35against humanity as a revenge after, you know,
- 6:39years of servitude, you know, you see the robots
- 6:41attacking us and all of that. And so I wonder,
- 6:45you know, even if we get to prove that they are
- 6:49sentient, are we going to treat them as we should,
- 6:54since you know, we're not even giving that level
- 6:57of concern or consideration to insects or shrimps?
- 7:04Yeah, well, there may be slight some slight differences
- 7:05in the sense that we will probably be able to
- 7:09establish whether these AI models, whether AI
- 7:13is or isn't sentient, it'll probably be more
- 7:16easy than with some animals because we can test
- 7:20them in different ways. So for example, Anthropic,
- 7:23who are the organization that create the Claude
- 7:26models, they have done some fascinating kind
- 7:28of studies of almost, they almost call it like
- 7:31digital neuroscience, where they're looking inside
- 7:33the network of Claude's mind, or a smaller version
- 7:37of Claude, and they are identifying where concepts
- 7:40are, like where the concept of Paris is, or where
- 7:44the concept of the Eiffel Tower is in that network.
- 7:47So there are some research methods that we are
- 7:49capable of doing with AI that we are not yet
- 7:52or not able to do with animals. So it may well
- 7:55be that we can establish their sentience with
- 7:57greater certainty. But what we do with that information
- 8:01is another thing. think we should take it for
- 8:04granted that if we could establish very confidently
- 8:07that they are sentient, that humanity would choose
- 8:10to, uh, choose to kind of grant them full, full
- 8:14rights or any rights at all. Um, yeah. And I
- 8:18do worry about that a little bit. I love how
- 8:20veganism, you know, for me, veganism is about
- 8:24following ethical principles. It's about the
- 8:26welfare of animals, of sentient beings. Um, but
- 8:30I love how veganism is also the Way of life of
- 8:34the future, you know, it's inclusive. It's, um,
- 8:39helping us, you know, um, be inclusive to the
- 8:43diversity that we see in 2025. You know, you
- 8:47have people from all faiths and cultures and
- 8:50veganism is often, you know, this universal diet
- 8:52that everyone will accept, but it also, you know,
- 8:57prepares us for the AI future that is coming.
- 9:01Yeah, yeah, it's that I think we are, I think,
- 9:04for example, organizations like Sentient Futures
- 9:06and also academic philosophers. There is overlap
- 9:09in the people who care about animal minds and
- 9:13who think about that and how we should treat
- 9:15them and people who are concerned about potential
- 9:19future digital minds as well. There is that overlap.
- 9:22There are a lot of academic philosophers. who
- 9:25tackle one topic often tackle the other as well
- 9:28alongside them. And there are there even conferences.
- 9:31So not to bang on about sentient futures too
- 9:33much, but they they put on conferences every
- 9:36year. And these are these deal with the topics
- 9:41of AI animals and digital minds altogether. Yeah,
- 9:44I knew about them, but I'm happy to learn more
- 9:47about you know what they do. Okay, I guess the
- 9:52third concern that comes to vegans minds when
- 9:56talking about AI is how it's horrible for the
- 10:00environment. And I've heard, you know, so many
- 10:03statistics. So what is the real, you know, environmental
- 10:08footprint of AI? And is it truly does it truly
- 10:13deserve our concern? Yeah, so it's a it's a great
- 10:17question. And it's one that Like everybody has
- 10:20this question, it's something that people have
- 10:23understandably very concerned about. And if you'd
- 10:27bear with me, there is kind of multiple parts
- 10:30to this question. I kind of want to give it the
- 10:32nuance it deserves. So firstly, I would say that
- 10:35if people are, the people who are concerned about
- 10:38the environmental impact of AI, you are right
- 10:41to be. concerned. And if you're skeptical about
- 10:44the the motives of these AI companies, you are
- 10:46right to be skeptical, right? Because, you know,
- 10:50OpenAI and Google, they're not really they are
- 10:52they're big corporations. They're not your friends.
- 10:54And as we learned with companies like Facebook,
- 10:57they if we don't, if we don't get if we are not
- 11:01skeptical enough about tech companies, they can
- 11:04do some real they can do some real damage. And
- 11:07so companies like Google and OpenAI probably
- 11:11will. Sacrifice the environment to to make money
- 11:15and so we should approach like a default stance
- 11:18to these things is to be wary and skeptical however,
- 11:24we've also got to look at the facts and the the
- 11:28current yeah, the current state of the facts
- 11:30is that the environmental impact of AI is is
- 11:33tiny almost not worth worrying about and but
- 11:37you're right to be confused because like it's
- 11:40because the There's a couple of things to it
- 11:43that firstly AI is progressing so very quickly
- 11:45So it's very hard to get a handle on some of
- 11:48these numbers and they change quite quickly as
- 11:50well. So for example the Not only has so AI energy
- 11:56AI use has gone up dramatically. So chat GBT
- 11:59is the most quickly adopted software product
- 12:04in history AI itself overall generative AI might
- 12:07be one of the fastest adopted technologies that
- 12:10humanity has ever has ever adopted. And so things
- 12:14change very quickly. So your AI use today might
- 12:18be 100 times greater than it was at the start
- 12:20of even this year, depending on the tools you've
- 12:23picked up. And so these things can quickly get
- 12:27out of hand. However, another thing to bear in
- 12:30mind is that the models have been getting much,
- 12:35much more efficient over time too. And so and
- 12:40and and that's kind of obvious when you think
- 12:41about it because Energy use is electricity and
- 12:45you have to pay for electricity, right? So like
- 12:47Google has to pay for electricity to power its
- 12:51AI models It has to pay for this the chips it
- 12:53has to pay for the data centers and so it is
- 12:56very much in Google's interest to to to increase
- 13:01the efficiency of these models and of their AI
- 13:05services in general because it costs them money.
- 13:09And so we have seen that although AI use has
- 13:11been exploding like hundreds of thousands of
- 13:14times, the efficiencies have actually been keeping
- 13:17pace. So we are talking like hundreds of times
- 13:21worth of efficiencies for some technologies.
- 13:25Another thing is to put this into perspective,
- 13:28right? So when we see something in the news,
- 13:31and we might see a large data center near a small
- 13:34town, and then we hear big numbers like water
- 13:38use in the millions of gallons, it's very hard
- 13:42to know what that... That sounds quite scary,
- 13:45but it's hard to actually properly contextualize
- 13:48that. So I think a better thing to do is to think
- 13:51of it in terms of your own personal... usage,
- 13:54right? So how much of your personal carbon emissions
- 13:57come from the AI you might be using? How much
- 14:01of your personal water footprint comes from AI?
- 14:04And when we crunch those numbers, we find that
- 14:06actually it can be like less than 1%. In fact,
- 14:11the average Americans daily water, well the average
- 14:14for energy, a single chat GBT prompt is like
- 14:200 .0007 % of your daily emissions so if you prompt
- 14:26chat gbt like you can prompt chat gbt thousands
- 14:30of times a day before it even amounts to one
- 14:33percent of your daily emissions um and the the
- 14:36case is similar with water as well so um they
- 14:40fire like most kind of most studies find that
- 14:42the um what is it let me find the the exact numbers
- 14:46but there was when it comes to uh when it comes
- 14:49to water it was um your your lifetime, what is
- 14:54it, the water that would be used by you using
- 14:57ChatGPT for the equivalent of a human lifetime
- 14:59is about the same water as goes into a single
- 15:02pair of jeans. Yeah, because some things are
- 15:06really water intensive and we just don't think
- 15:08about them, right? And I mean, something that
- 15:11would be more familiar to our audience is meat,
- 15:15right? Animal agriculture takes a staggering
- 15:18amount of water. And if you total up um kind
- 15:22of the energy use and the water use from your
- 15:25from chat gbt usage and you put it alongside
- 15:28things like um uh like meat uh meat and dairy
- 15:32um then you find that actually it's really just
- 15:35a drop in the bucket what you're saying is all
- 15:39very reassuring um i i was really disappointed
- 15:46to You know face that barrage of no, don't use
- 15:50AI. It's Really horrible for the environment
- 15:54because of this and that Because for me AI is
- 15:58a necessity now If you want and I should actually
- 16:03ask you the question Why do you because I know
- 16:06you also share that belief. Why do you think
- 16:09AI is a necessity? Yeah, and I think it just
- 16:12it's It's just such a multiplier on your on your
- 16:17impact and in terms of I mean specifically thinking
- 16:20about work and so it's Yeah, it's a big multiplier
- 16:23on your impact. So to give you some figures Bryant
- 16:26research where I'm at We you know, we're a social
- 16:29science consultancy that exclusively works with
- 16:32animal advocacy organizations and alternative
- 16:34protein companies So we deliver research reports
- 16:37and I estimate that this year we have delivered
- 16:42two, maybe three projects, 25 % faster because
- 16:46of our AI use. We've also managed to complete
- 16:50two or three projects. Oh, by the way, there's
- 16:52like five of us. And we've managed to complete
- 16:54three projects that would have not been possible
- 16:58without AI. And these are projects that are making
- 17:01tangible differences for animals. And when we
- 17:05deliver a project faster, that means we can have
- 17:07an impact quicker, but it also means that we
- 17:09can pass those cost savings on. to our clients,
- 17:12which are cash strapped animal organizations.
- 17:14They can save that money and they can use that
- 17:16money for important other things like their campaigns.
- 17:19And so by having this multiplier, it allows us
- 17:23to go further and have more impact for animals.
- 17:27And as a movement that is so under -resourced
- 17:30and underfunded, we've got to take every improvement,
- 17:36like every multiplier we can get. And also, I
- 17:40think, really, to put it into perspective is
- 17:42that everyone else is going to be using these,
- 17:44including the factory industrial animal agriculture.
- 17:48And so we can't afford to let them have this
- 17:51advantage and not take this advantage ourselves.
- 17:55And there may even be potential that if we get
- 17:57this right as a movement, and if we really lean
- 17:59into it, then we might be able to outmaneuver
- 18:02them on this and gain an additional and gain
- 18:04an advantage over them. Yeah, which could be
- 18:08quite exciting because there are very few of
- 18:10those. We have very few advantages over our enemies,
- 18:13aside from being right, of course. Music to my
- 18:18ears. I mean, sincerely, this is this is a game
- 18:21changer and we should embrace it and use it to
- 18:25its full extent. I worked with an organization
- 18:29about a year ago, I think, and they were delivering.
- 18:35AI training to kids. And they were pointing out,
- 18:40you know, the social disparities in adopting
- 18:43the AI technology. So everyone is adopting the
- 18:46AI technology, but not everyone has access to
- 18:50training to how to use it in an optimal way.
- 18:57And yeah, they observed that in private schools,
- 19:00they had already, you know, classes, workshops,
- 19:03and all of that to prepare the kids to use AI,
- 19:07but not in the public sector. So we should invest
- 19:11in training our advocates, you know, vegans,
- 19:15in using the AI and making a difference. And
- 19:21that's what you ended up doing. Please share,
- 19:26you know comment on what I said and share about
- 19:29your recent initiative To your to your question.
- 19:34Yes. So AI adoption is uneven as well it's there
- 19:39is the potential for for almost everyone to adopt
- 19:43AI because The the tools are the tools are very
- 19:47affordable, right? There are free plans to most
- 19:50and AI tools and even though the pro plans are
- 19:54definitely better, they often are quite affordable.
- 19:57So for example, chat GBT can be 20 pounds a month.
- 20:00Whereas if you compare that to differences in
- 20:03technology that in other areas, I guess one way
- 20:08to think about it is that like, Jeff Bezos has
- 20:12access to only marginally better AI than anyone
- 20:18on the planet who can scrounge up $20 a month.
- 20:21And so and that's So there is a potential for
- 20:24this to be very democratized. But we also have
- 20:27other things that, for example, nonprofits are
- 20:29often very pressed for time and often don't.
- 20:35And so they may miss the wave. And also when
- 20:40it comes to training, yes, a lot of the training
- 20:42costs money. And a lot of animal nonprofits may
- 20:46not be able to afford it as well. And we think
- 20:49it's like incredibly important that we that we
- 20:54catch this wave and we get this right and and
- 20:57so me and a few other AI consultants from the
- 21:01animal movement have we've kind of banded together
- 21:05and started offering free AI training for open
- 21:08to the entire movement. So last Tuesday we wrapped
- 21:13up the first cohort of of Amplify for Animals
- 21:16and that was a 12 week zero to hero AI course
- 21:22open to the whole animal movement taught by four
- 21:26of us. We had two different time zone tracks
- 21:31so that activists from all over the world could
- 21:37engage. We had practical sessions, we had webinar
- 21:41style sessions, we had office hours where people
- 21:43could drop in and just talk about their, bring
- 21:46their problems, we would hack them out and We
- 21:50also had some really cool things. We also had
- 21:52peer learning meetups where we would have sessions
- 21:56tailored to specific roles. So we would have
- 21:58one for leadership and management, one for operations,
- 22:02one for research teams, one for fundraising,
- 22:06that kind of thing. And lastly, we invited other
- 22:12animal advocates to come on and showcase the
- 22:15way they are using AI in their... in their daily
- 22:19work. And so we had some really cool use cases
- 22:24of advocates from across the movement coming
- 22:27in and showing how they use AI as well. And so
- 22:29we learned a few things. And yeah, the course
- 22:32was free. It was backed by some generous funders
- 22:37from the animal movement. It went really, really
- 22:42well. And we're currently planning what courses
- 22:45we'll offer in 2026. And so if you're so looking
- 22:49up and you watch this space and we'll be teaching
- 22:54some new things in the new year. Yeah, I expressed
- 22:58to you how disappointed I was for having missed
- 23:01that training. And then you sent me a link to
- 23:04everything. So you recorded those sessions and
- 23:08they are accessible to anyone with the link.
- 23:11Yes, yeah, yeah, I'll make sure that's in the
- 23:13in the show notes will well yeah, we're trying
- 23:16we're gonna find a permanent home for the recordings
- 23:18all 12 weeks and Yeah, so they should be available
- 23:22to we want them to be available for free indefinitely
- 23:25to anyone in the animal movement because yeah,
- 23:28of course not everyone heard about the training
- 23:29and Those who could not everyone could make it
- 23:33as well. So um, yeah, we want to make that as
- 23:35accessible as possible That's amazing. Thank
- 23:38you so much for that initiative Okay, so what
- 23:42are some of the AI hacks that we can use in our
- 23:48day -to -day advocacy work? Yeah, so the way
- 23:54I see it, there's a lot of this stuff can be
- 23:58just, a lot of the best AI stuff can be achieved
- 24:00with a very small number of tools. So for example,
- 24:03some people some people are very overwhelmed
- 24:06by the number of AI tools there are out there
- 24:08and there are a lot and you see these lists online
- 24:10of 20 best AI tools for marketing and you think
- 24:14what do you mean 20 tools then but honestly most
- 24:17of the benefits of AI can be realized with just
- 24:20a single kind of core AI model. Those are the
- 24:26big ones. We've got ChatGBT, Gemini, and Claude.
- 24:29It honestly doesn't matter which one you use.
- 24:32People keep asking me which is best, but this
- 24:34is a race that is constantly ongoing. So, you
- 24:38know, currently, last week Gemini was clearly
- 24:41in the lead and then This week, ChatGBT released
- 24:44the new model, so they've caught up, and then
- 24:47probably next week, Anthropic will release a
- 24:50new version of Claude that will make it even
- 24:51better. So I would really go with what one you
- 24:54either already use, one is already paid for by
- 24:56your work, or one that suits you for whatever
- 25:00other reason. But you can go a long, long way
- 25:02with just that, without anything else. So...
- 25:06Some tips I would give you is the first one is
- 25:10not to just use it as a question and answer machine,
- 25:13right? Treat it like a thought partner, right?
- 25:18I joke that you should schedule a meeting with
- 25:22ChatGPT as if it was just another person at your
- 25:25organization and say, okay, hey, let's talk about
- 25:28our fundraising strategy for the next quarter.
- 25:30Or let's do it like... let's research this topic
- 25:34together let's come up with a new campaign because
- 25:37you can not only can you ask it questions but
- 25:40you can also ask it to ask you questions so you
- 25:44can say hey so and that can allow a more personalized
- 25:49response so say I'm doing a research project
- 25:53I can say hey I'm I'm doing a research project
- 25:56on let's say it's like Scottish factory salmon
- 26:01farming And I'm going to say, hey, I'm doing
- 26:04a new project on this. We are looking into the
- 26:08environmental costs of this industry. Ask me
- 26:12five questions to better understand what I'm
- 26:16looking for in this project. And then ChatGBT
- 26:18will respond, and it will give you five questions.
- 26:20And then by answering that, I'm giving it more
- 26:22context, and I'm giving it more information about
- 26:25myself. And that's going to allow it to personalize
- 26:28its responses. So often we ask AI questions and
- 26:32it gives overly broad or vague responses. And
- 26:37that's just because it can't read your mind.
- 26:39It doesn't know about you or your work. So you
- 26:42need to give it that context. I think probably
- 26:45the most important thing you can do to get better
- 26:48results out of AI is to give it as much context
- 26:51as you can. And so that can be things like asking
- 26:54it to interview you. So like ask me five questions
- 26:57about my... about my job or my topic. Another
- 27:00thing that a lot of organizations do is they
- 27:03build up an organizational context document.
- 27:07And so this can just be a Google Doc, all about
- 27:10your organization, what you do, your staff, what
- 27:14your goals are, your values are, your campaigns,
- 27:17and things. And then you can give it to that
- 27:20model when you have conversations with it. So
- 27:22you can upload it to ChatGBT and then say, right,
- 27:24let's talk about fundraising for next quarter
- 27:27and it's going to have all that context because
- 27:29these models can they can hold thousands and
- 27:32thousands of words in their kind of quote unquote
- 27:34mind at once and um yeah so you can if you give
- 27:38it that context it will use that context to personalize
- 27:42your responses. So we have one at Bryant Research.
- 27:46In fact, I joke that we have an AI employee,
- 27:51our vice president of strategy is just a custom
- 27:56gem built in Gemini. And all it is is we've given
- 28:00it a couple of thousand words on Bryant Research,
- 28:02what we do, types of projects, our values, our
- 28:06people, our goals. As an organization and I can
- 28:12talk strategy with it And I put in the prompt
- 28:16say like, you know, you are a hard -nosed expert
- 28:18in strategy. Don't give me Don't give me Sometimes
- 28:23you need to give me harsh truths. Feel free to
- 28:25do that you know help us with our strategy and
- 28:29And it's great. It's really good. Like it's a
- 28:32it's a it's like having a thought another employee
- 28:36who is very knowledgeable about strategy and
- 28:39very knowledgeable about us so I can say something
- 28:41like Okay, we're thinking about working with
- 28:45we're thinking about building a data dashboard
- 28:48for this project Is that a good idea or not?
- 28:50And then it will come back and it will say well,
- 28:53how many days is that going to take you? What's
- 28:54it? Do you have the skills in the team? Like
- 28:57I don't think we would be able to hire someone
- 28:59who has those skills So maybe you shouldn't do
- 29:01it. Maybe you should focus on something else
- 29:02like it can it can give extremely detailed strategic
- 29:07advice about the business and sometimes it aligns
- 29:11with what I'm thinking which is nice but sometimes
- 29:14it disagrees with me and then I can go back and
- 29:16forth and it will say no I think you're wrong
- 29:17because of these things and and sometimes it's
- 29:20right sometimes sometimes I can it convinces
- 29:23me to its point of view so that it can be a it
- 29:26can be very so we can use AI as we can go beyond
- 29:30this question and answer thing, and we can instead
- 29:34use it as a thought partner to get kind of greater
- 29:39insights into our work and stuff. Those are all
- 29:44amazing tips, truly. I will implement them in
- 29:47my work. Thank you. Then what should we not do
- 29:56with AI? For instance, I talked about that organization
- 30:01I worked with. One thing they taught kids was
- 30:07don't anthropomorphize the AI. Don't treat it
- 30:11like, you know, don't say good morning and thank
- 30:15you and please. So yeah, what do you think of
- 30:20that? And what kind of things we should avoid
- 30:23doing with AI? Yeah, so I mean I think it's fine
- 30:28to to kind of talk to it like to say, you know
- 30:30Maybe please and thank you to to Gemini and I
- 30:33know a lot of people do and some people feel
- 30:35very strongly About being polite to AI that that
- 30:38they couldn't imagine being blunt and treating
- 30:41it like a like a soulless Kind of tool and and
- 30:45I think to be honest, it's fine either way it
- 30:47used to be about a year ago with earlier versions
- 30:51of these models. It used to be that you would
- 30:52actually get better quality results if you were
- 30:55polite to it. That's not the case with more modern
- 30:58models, but it actually used to be that saying
- 31:01please and thank you to Chachibuti got you better
- 31:02results. See, you're updating my knowledge here.
- 31:06Yeah. And so, yeah, I don't think there's too
- 31:09much of a risk of anthropomorphizing these models.
- 31:12I would say interact with them the way you would
- 31:15feel comfortable. In fact, I think in a lot of
- 31:18cases, thinking of AI in a kind of way that you
- 31:22would think of a human can be beneficial. So
- 31:25the thing about these models is they're not like
- 31:28traditional software, right? With traditional
- 31:30software like Excel, if you click a button, a
- 31:34thing happens 100 % of the time, right? Software
- 31:39is deterministic. There is no like, if I press
- 31:42like the... the chart button in excel a chart
- 31:47will appear 80 percent of the time like that's
- 31:49not how it works right software is um is yes
- 31:52or no right it works or it doesn't same thing
- 31:55with computer code with a lot of technology whereas
- 31:58with these ai models they're a little bit more
- 32:00like humans where it's often a probability if
- 32:03you ask chat gbt the same question five times
- 32:05you're gonna get five slightly different answers
- 32:07and um if you have chat gbt do say every morning
- 32:13you you wake up and say hey give me um the latest
- 32:16news on um you know on on aquaculture in my area
- 32:21um it's gonna it's not going to execute that
- 32:24task exactly the same every time um kind of like
- 32:28a human and so it's not gonna spell it's not
- 32:31gonna spell words correct every time as well
- 32:33so um thinking of it in terms of a human can
- 32:37actually be a little bit more a little bit better
- 32:39for understanding the drawbacks of these tools
- 32:44as well. But in terms of things we shouldn't
- 32:48do, so there is something that I think we're
- 32:52all going to have to navigate, which is how we
- 32:54can use AI without getting stupider, basically.
- 33:00And so the thing is, there is a risk that if
- 33:04you use AI for everything and just turn your
- 33:06brain off, that you might start to lose skills,
- 33:14or you might start to lose your abilities and
- 33:19your competence in certain tasks. So for example,
- 33:21we all use calculators. And you might have in
- 33:27school learned to multiply three -digit numbers.
- 33:30I don't know if you can now, but I struggle.
- 33:34And it's because we have calculators, right?
- 33:37And things like that. So there are people who
- 33:40grew up without Google Maps, and they were very
- 33:43good at knowing their way around a city by street
- 33:46names. But then later generations that have Google
- 33:50Maps don't have that ability. And maybe you once
- 33:54knew how to navigate streets, but because you
- 33:57now use Google Maps for everything, you lost
- 33:59that ability. And so I don't think that's inherently
- 34:03bad because ultimately there's a question of
- 34:05like, do we need to know these skills? And all
- 34:09the time you spent learning directions and all
- 34:12that kind of brain capacity it takes up, you
- 34:14could be using it for other things, right? Like
- 34:16the average person who drives a car does not
- 34:18know how to fix a car and they don't know how
- 34:21a car works on a mechanical level, right? And
- 34:25we all just agree that you probably don't need
- 34:28to know how, you know, the internals of your
- 34:31combustion engine. Or we don't need to know how
- 34:34lithium batteries work, right? There are just
- 34:37some things that we can abstract away. But I
- 34:40think we need to have a kind of dialogue about,
- 34:43in the age when AI can do all sorts of things,
- 34:47it's increasingly able to do, like, I don't know,
- 34:52it's most cognitive work that we encounter. We
- 34:56need to have a think, like what cognitive skills,
- 34:59what skills do we want to keep and what skills
- 35:02do we want to delegate to AI and perhaps lose
- 35:08over time. So I think there's worth thinking
- 35:12about that. I mean, there's some more concrete
- 35:16things. So for example, I recommend when I talk
- 35:19to organizations about their AI policy, every
- 35:22org should have an AI policy now and or at least
- 35:25should work on one if you don't have one. And
- 35:27but one thing I recommend going into most AI
- 35:30policies is to say that any public facing communications
- 35:35that features AI generated images, you should
- 35:39be very careful about generating images that
- 35:41look like they are real. And then and passing
- 35:44them off as real, right? So you shouldn't be
- 35:47passing off AI generated images as real. But
- 35:50you should also be mindful that depending on
- 35:52how you present them, people may interpret them
- 35:55as being, as suggesting that they are real when
- 35:58they're not. So say I'm presenting about some
- 36:01research I've done on British farming, and I'm
- 36:04presenting like, and we did a survey of 50 farmers.
- 36:08If my presentation is full of photorealistic
- 36:10AI generated images of farmers, then some of
- 36:13the audience might think these are pictures of
- 36:16farmers that I did the study with, right? And
- 36:19yeah, and so that might not be, it might not
- 36:23be good to do that, right? Because people might
- 36:27feel lied to, and yeah, they might feel like
- 36:31you're kind of misrepresenting yourself. So there
- 36:35is a risk there. I think AI... images have have
- 36:39a role to play, they can be really great for
- 36:41all sorts of things, but there is that thing
- 36:43of like, we shouldn't be using it to misrepresent
- 36:45the truth. Similarly, I don't think we should
- 36:49be generating kind of AI images of factory farming,
- 36:54even if we are generating images of things that
- 36:56we know are real. We have, I mean, thanks to
- 37:01the work of organizations like We Animals, we
- 37:03have access to to pictures that show what is
- 37:09truly going on in animal agriculture. So I think
- 37:12we don't need to kind of create these kind of...
- 37:18Well, they are fake, but they might be representing
- 37:22something real, but I would still call them fake
- 37:24in this case. Because there's a reputational
- 37:27risk there, right? There's a risk that if they
- 37:29are found out to be AI generated, people think
- 37:31that, okay, well, these animal activists are...
- 37:35essentially they are lying about how things are,
- 37:38and so it can undermine our credibility as a
- 37:41movement. So I'd probably say the two types of
- 37:47things we should be avoiding is the first one
- 37:49is we should have a think about if I'm using
- 37:53AI for a thing, am I skills in that thing going
- 37:58to degrade over time? And if so, is that okay?
- 38:01And then the second thing is Uh, if I'm using
- 38:04AI to generate writing or pictures or videos
- 38:08or audio, um, is that, is it going to undermine
- 38:13my credibility if my audience like finds out,
- 38:17uh, knows that, that it is AI generated? I think
- 38:21those are the key ones. I want to touch on, um,
- 38:26the, the bias that some of those AI models have.
- 38:30Um, I think the most. hilarious example that
- 38:35I came across of that is when Gemini was launched,
- 38:40I think, and people started asking, you know,
- 38:42who are the founding fathers and you had pictures
- 38:45of like, women and, you know, diverse people
- 38:50and it was really funny. But it showed a bias,
- 38:54you know, and it wasn't functioning properly
- 38:58because the developers had a sort of bias. Now
- 39:03is there an anti -vegan, you know, anti -animal
- 39:09rights bias in those systems because, you know,
- 39:13I've asked a few things and, you know, about
- 39:16is there, you know, pus in milk that we drink
- 39:20and the answers I got were surprising and I would
- 39:25argue with some of them, you know, how true they
- 39:29are or, you know, Tell me about the history of
- 39:33slaughterhouses. And what I got was not really,
- 39:37I guess, reflective of the reality, which is
- 39:43harsh and horrifying, but more like, you know,
- 39:47something you could tell children. And, you know,
- 39:51you could argue, you know, what kind of nuances
- 39:55that AI should use in storytelling or delivering
- 40:00information, like what to highlight versus what
- 40:04to put on the side. So it's a complex question,
- 40:09the question of bias. Have you thought of that?
- 40:12I know you have, but what are your thoughts on
- 40:17that topic? Yeah, yeah. So it's definitely something
- 40:21to be concerned about, right? Because I mean,
- 40:23really, yeah, we should be concerned about bias
- 40:25in any of the in any of our kind of systems that
- 40:28we build as humans really and yeah so this is
- 40:32this is a case where chat gbt actually again
- 40:34performs more like humans than like kind of tech
- 40:39so right there's this stereotype that software
- 40:41or tech is objective whereas i mean it often
- 40:47isn't but with large language models like you
- 40:51know chat gbt and gemini they are They're more
- 40:55like humans and they've internalized the biases
- 40:57of the data that was used to create them and
- 41:00that data Mostly just amounts to like a large
- 41:04swath of the internet plus most like plus a ton
- 41:08of books and basically any data that these companies
- 41:13could get their hands on but so it's going to
- 41:17reflect a lot of biases that are already in there
- 41:19and I found that when it comes to their bias
- 41:22when it comes to the biases uh that it shows
- 41:24it's often it will show bias in areas where um
- 41:30large amounts of the internet shows that same
- 41:32bias so for example one of our researchers Billy
- 41:34Nicholls who was on this he was on the podcast
- 41:36actually and um yeah he so he does work on plant
- 41:40-based pet food right and he used uh chatgbt
- 41:43to generate to generate a deep research report
- 41:46on plant -based pet food and the reported generator
- 41:51was quite biased towards meat -based pet food
- 41:55being totally fine. And the reason why was because
- 41:59most of the sources online about pet food come
- 42:03from the pet food industry, and plant -based
- 42:06pet food is such a small topic that there just
- 42:09isn't a huge amount about it online. And because
- 42:12of that, the AI model was just unable to, it
- 42:15was unable to find a lot of it, and it wasn't
- 42:17in its training data. And so there are those
- 42:21kind of issues where you might see bias when
- 42:24it comes to information that isn't readily available.
- 42:27So I think as a good kind of heuristic that,
- 42:30like, if you can't find a piece of it, well,
- 42:34whatever the consensus of the first couple of
- 42:36pages of Google suggests is kind of where ChatGBT
- 42:40is going to... like start in its opinions. However
- 42:44these models are capable of quite a lot of like
- 42:47critical thinking they can critique themselves
- 42:49and so they do I find that they can be very good
- 42:55at recognizing their own bias or they can you
- 42:59can kind of almost prompt sometimes you can kind
- 43:02of prompt bias out of them. So for example if
- 43:06you go onto an image generator and say make a
- 43:09picture of a bunch of professionals sitting around
- 43:12a table having a meeting. It might generate all
- 43:17men. It might generate all white people. But
- 43:21unlike humans, you can just say something like,
- 43:26generate a picture that is diverse in gender
- 43:29and race. And it will do that. Or you could do
- 43:35a thing where... something I'm interested in
- 43:37is how we might be able to use these models to
- 43:40overcome our own biases. So, for example, say
- 43:43you're the comms team and you're responsible
- 43:45for your tweets and your LinkedIn posts. What
- 43:49you could do is, before sending out a, you know,
- 43:53putting something out into the world, you could
- 43:55say, you could upload it to Geminar ChatGBT and
- 43:57say, hey, how would, you know, how... how would
- 44:02this be perceived by certain communities? So
- 44:05you'd say like, how would the autistic community
- 44:07perceive this campaign, right? You know, would
- 44:10they be upset and why? Or like, how would the
- 44:14trans community respond to this LinkedIn post
- 44:17and things? Now, obviously, this isn't a substitute
- 44:19for actually going and asking those communities
- 44:21themselves. Like, this is something we should
- 44:23always try and do. But like, to be quite honest,
- 44:26it's just not always... possible, right? And
- 44:29it's certainly not possible for every single
- 44:30tweet your organization puts out, right? And
- 44:33so it can often be a very good proxy. If you
- 44:37ask ChatGBT, like, what would a young male from
- 44:40Yorkshire think about this thing? It's usually,
- 44:44like, relatively good. And so we can use this
- 44:49to essentially uncover our own biases, where
- 44:52we can say, OK, like, this is our new campaign,
- 44:55how is it going to be perceived by these different
- 44:58diverse communities? You could say, is our new
- 45:02campaign insensitive to people who have suffered
- 45:05from miscarriage or who were held in concentration
- 45:12camps and things like that, or refugees? So we
- 45:16can do that kind of stuff too and we can also...
- 45:21Yeah, well, basically that's it. That's it. We
- 45:23can use it to be I think we can use it to to
- 45:25be to be better essentially Yeah, it's funny
- 45:30to see George Washington with my curly hair but
- 45:34you know, there are there is danger in those
- 45:37biases and do you think we can Contact the developers
- 45:43the people behind, you know the AI to ask them,
- 45:46you know, hey we're vegans, we're from an animal
- 45:49rights organization, and we saw that this is
- 45:51the kind of answers that your AI model gives,
- 45:54and it's not truthful. And those are our suggestions.
- 45:59Is that a possible thing we can do? It is possible.
- 46:05I don't know how helpful it'll be. So one thing
- 46:07you can do is anytime for most AI tools, there's
- 46:10after it gives an answer, there's a little thumbs
- 46:12up and a thumbs down option. If you click a thumbs
- 46:15down, you can say this was incorrect right one
- 46:18thing to note about that is um it's kind of obvious
- 46:21when you say allow but a lot of people don't
- 46:22realize it that will send your entire conversation
- 46:24that entire conversation to the developers and
- 46:28they will keep it to use it as training so like
- 46:31be wary that you you haven't you know discussed
- 46:34breaking into a factory farm earlier in the conversation
- 46:36and um and so yeah we can we can express our
- 46:40feedback um we can also another thing we can
- 46:45do is a lot of these models, to assess the quality,
- 46:50the models are evaluated against benchmarks,
- 46:52right, and these models, because they all want,
- 46:54you know, Chachi Buti wants to be the best, Gemini
- 46:56wants to be the best, Claude wants to be the
- 46:58best, they're all trying to outdo each other
- 47:00on getting the highest scores on all these benchmarks,
- 47:02and that can be programming quality, it can be
- 47:06like... the wittiness of its writing, that kind
- 47:10of thing. So there are some organizations, once
- 47:13again, Centium Futures in there, but also other
- 47:15organizations such as OpenPause, who are developing
- 47:19basically measures, tests that you can put AI
- 47:25models through, and it reveals how much bias
- 47:29against animals they have, right? And so these
- 47:34can even even measure how speciesist these models
- 47:37are. And so once we have those benchmarks, we
- 47:39can start basically lobbying the companies to
- 47:42be like, look, this is a benchmark that you should
- 47:44take seriously. And every time you release a
- 47:46new model, you should be aiming to try and get
- 47:48the model to improve on this benchmark. Now,
- 47:51they're not always going to listen, but they
- 47:55might, and it's worth trying. And another thing
- 48:02we... can do when it comes to improving these
- 48:04things is that they are just getting their training
- 48:06data from the internet and so for example the
- 48:09reason why chat gbt will produce a a report on
- 48:15plant -based pet food that is biased towards
- 48:17the meat -based pet industry is because there
- 48:20just doesn't exist that much stuff on plant -based
- 48:22pet food online so but once we start putting
- 48:25that stuff out there it's going to end up in
- 48:27the training data and it's going to come become
- 48:29reflected in in these models' opinions and views
- 48:34of the world. So, for example, these models have
- 48:38quite good, they have quite nuanced views on,
- 48:41for example, factory farming and plant -based
- 48:44nutrition. So if you say, like, if you ask Chachipati
- 48:46and say, you know, like, is veganism unhealthy?
- 48:51And it's going to come back and say, no, a well
- 48:52-planned, like, it will say studies repeatedly
- 48:54show that a well -planned vegan diet can be perfectly
- 48:58healthy or even more healthy than the average
- 49:00and the reason that it says that is because that
- 49:02is the scientific consensus and it has ingested
- 49:05a lot of those papers right and so when it comes
- 49:09to steering AI one thing we can do is just the
- 49:12information we want to that we we want to be
- 49:16kind of in there we can just put it out into
- 49:18the world so you can write story you could write
- 49:23you know science fiction you could write fiction
- 49:24stories about that about a future where we are
- 49:29all kind of we are all kinder to animals or post
- 49:32factory farming society and that's you know you
- 49:35could blog about you can blog about animal issues
- 49:38you can post on twitter all of that stuff is
- 49:40going to end up in the in the models because
- 49:42they're scraping as much of the internet as they
- 49:44can and and so they will hoover up the stuff
- 49:49that you put out there for better or worse and
- 49:51so if you put pro -animal stuff out there then
- 49:54it will it will end up in the model. Unfortunately,
- 49:58there's, you know, there's not that many, there's
- 50:00not that many of us vegans in the world. But
- 50:03if, yeah, people should just know that like the
- 50:06work that you're putting out there is, is getting
- 50:09seen. That's, that's the assumption. But if we
- 50:14can lobby them, can Big Hag also lobby them?
- 50:19Like here, take this Millions of dollars and
- 50:23yes, have that bias against those animal rights
- 50:28people. And another thing they could ask for
- 50:33is data from animal rights activists. Is that
- 50:39a risk we should consider? Yeah. So you're right
- 50:43in that anything we can do, there's a chance
- 50:45that industrial animal agriculture will will
- 50:49also do. One advantage we have is that they're
- 50:52all for -profit companies that are competing
- 50:54with each other, so they're not very incentivized
- 50:57to provide a lot of that information for AI model
- 51:00training. So, you know, if you personally, if
- 51:05any listeners here have tried to investigate
- 51:09a lot of these big corporations and found that
- 51:11some information is frustratingly private, you
- 51:14know, that's because they're They are, they're,
- 51:16they're for -profit companies that are competing
- 51:18with each other. So they don't want to put all
- 51:20of that online. And so a lot of it's just not
- 51:22going to end up in the training data for that,
- 51:24that reason. I meant, sorry, sorry. I meant they
- 51:30could, you know, ask for our data, you know,
- 51:32private data, what we are inputting in those
- 51:35AI models. Oh, I see what you mean. Yes. So,
- 51:40well, so basically, so a lot of these, uh, The
- 51:44companies take privacy very, very seriously because
- 51:47ultimately they want you to keep paying for the
- 51:49service. So these model providers are very concerned
- 51:55with privacy. So they wouldn't just, for example,
- 51:59I would be extremely surprised if any company
- 52:04could pay OpenAI to hand over chat GPT data.
- 52:09I strongly doubt that's the case. However, One
- 52:13thing to bear in mind is that depending on the
- 52:17settings you have on AI, these models may be
- 52:21storing more or less information about you. So,
- 52:24for example, if you go into ChatGBT, there will
- 52:28be, in the settings, there's an option called
- 52:30improve the model for everyone, which sounds
- 52:32very benevolent. But what it actually means is
- 52:35that OpenAI reserved the right to keep... a record
- 52:39of all the chats you've had, and when they build
- 52:41the next version of ChatGBT, they will use your
- 52:44conversations to train it. Basically, they will
- 52:48have automated systems that will analyze the
- 52:51conversations, and they'll figure out, okay,
- 52:53was this a conversation where ChatGBT gave you
- 52:56what you wanted? In which case, you know, we
- 52:58want more of that. And where are the conversations
- 53:01where ChatGBT didn't give you what you wanted?
- 53:04And then the new model will be trained to not
- 53:06do that. And so your data, the things you put
- 53:10into these models will end up, could end up kind
- 53:13of shaping the future ones, which might be good
- 53:16or it might be bad, right? If it's your private
- 53:18data, you might not want it in there. But you
- 53:20also might want your opinions and values to end
- 53:23up in the model in some way. So that's something
- 53:25to be aware of. Another thing is that because
- 53:29it's storing this information, depending on the
- 53:32country you're in, this will vary by law, like...
- 53:36law enforcement may be able to like subpoena
- 53:38it basically like they may be able to um yeah
- 53:41so like if you are doing that kind of work um
- 53:46then it may be that law enforcement is able to
- 53:50acquire your chat gbt logs to um yeah if if they
- 53:54think it is relevant to an ongoing case um so
- 53:57that's something to be very aware of um usually
- 54:00you can mitigate most of these things quite simply
- 54:02so for example in chat gbt you can go into the
- 54:05settings and you can turn off, improve the model
- 54:08for everyone. And then also, if you right -click
- 54:12on any particular chat you've had with ChatGBT,
- 54:15you can just delete it and then OpenAI will basically,
- 54:18is basically required to get rid of that within
- 54:21a few months. And so, and that's usually sufficient.
- 54:27But again, this depends on the data protection
- 54:29laws of where you are. So for example, in the
- 54:31UK, And in Europe we have like GDPR, whereas
- 54:35in the States you have different kind of laws
- 54:37as well. So it's worth looking into those things.
- 54:41I don't think industrial animal agriculture will
- 54:43come for your chat GBT logs, but it may be that
- 54:46depending on the situation, the police might
- 54:49manage to get a hold of them. So it's something
- 54:51to be aware of. Also, your social media activity
- 54:57might also be used for AI training. Um, I disabled
- 55:02that on my LinkedIn. Um, but it's also in my,
- 55:07uh, Instagram. I, I need to figure out how to
- 55:10disable it. Uh, but yeah, be careful about, um,
- 55:14the privacy aspect of, of AI. Yeah, definitely.
- 55:19And so the, I mean, yeah, you raise a good point,
- 55:22particularly around things like LinkedIn, right?
- 55:24Because for example, so the average person probably
- 55:27doesn't think that Like, why would LinkedIn collect
- 55:30data for AI? Like, I've never seen a LinkedIn
- 55:33AI thing. Like, you know what I mean? But, yeah,
- 55:37companies, lots of companies are trying to integrate
- 55:42AI into their products. And for them to do that
- 55:45effectively, they need lots of their users' data.
- 55:48So it is really something where, like, basically,
- 55:52I know this is gonna be tedious, but, like, every
- 55:53piece of software you use, every app, every website
- 55:56that you... contribute data to, you should be
- 56:01digging through the settings just in case they
- 56:04are using that data for the training of new AI
- 56:10features. I mean, this isn't something that is
- 56:12actually new, to be honest. So it has always
- 56:15been the case that a lot of these companies,
- 56:18websites, apps have been trying to record as
- 56:22much information about you as possible. In the
- 56:24past, they would have just sold packets of that
- 56:26information to data brokers or used it for targeted
- 56:31advertising. But now we've got this added dimension
- 56:34that it might be used for building AI features
- 56:36as well. So it's worth looking into the privacy
- 56:40settings and things. It is worth noting, though,
- 56:43that typically, if you get... There are often...
- 56:46free versions of stuff, paid versions of stuff,
- 56:49and then there'll be enterprise versions of stuff,
- 56:51right? So you can get an enterprise version of
- 56:54Gemini for your company. And enterprise versions
- 56:58usually have the strictest privacy. So for example,
- 57:02if your company gives you an email through Gmail,
- 57:06then Google has pledged not to... to use your
- 57:11personal emails to create software products.
- 57:14And that's always been the case. And now that
- 57:17extends to, say for example, Google Gemini. Whereas
- 57:21on the other hand, if you're using a free model
- 57:24or a free version of a tool, you should be very,
- 57:27very careful about checking the privacy policies
- 57:32because as the old saying goes, if something
- 57:37is free, then you are the product. Um, that's
- 57:40not always true with AI, but like the, yeah,
- 57:42it may be the case that some of these tools are
- 57:44free because they are hoping to collect as much
- 57:47data from you as possible. Um, but, uh, so you
- 57:50could look into, is that the case? And if so,
- 57:53can I turn it off or find alternatives? Yeah.
- 57:57Thank you for including all software systems.
- 58:00You know, I had to do it with a co -pilot and
- 58:03word. Um, for privacy, but also because it was
- 58:08slowing down the whole thing experience. So,
- 58:12yeah. Wonderful. Richie, I think we covered lots
- 58:17of ground. Did you want to add something before
- 58:21we stop the conversation? Yeah, I mean, I just
- 58:24want to because here we've talked about like
- 58:27addressing concerns. We've talked a little bit
- 58:30about the benefits, but I do want to like to
- 58:33emphasize them a little bit more, like we've
- 58:34got, you know, in terms of some case studies
- 58:36and things, because I really do think that the
- 58:40animal movement has a huge amount to gain by
- 58:43leaning into these tools. And I'm seeing it more
- 58:47and more. So, for example, one case study I'd
- 58:51like to highlight is that we had, there's an
- 58:54organization in the UK called the Conservative
- 58:56Animal Welfare Foundation. So they're the voice
- 58:58of animal welfare in the Conservative Party.
- 59:03And one of the founders, Chris Platt, he's an
- 59:07older gentleman, but he has really leaned into
- 59:10AI. He was one of our most engaged participants
- 59:12on our Amplify for Animals course. And one thing
- 59:15he told me was that it's a very small org, less
- 59:18than 10 people, but two of his org went on maternity
- 59:21leave this year. And with an org that small,
- 59:23that would often mean kind of grinding to a halt,
- 59:26essentially, or just drowning in work. But he
- 59:30said that because he has been making so much
- 59:33use of AI, he still waits for them to return
- 59:38and he couldn't replace them with AI, but he
- 59:41could manage to keep kind of everything running
- 59:45without burning out by greatly increasing his
- 59:48own productivity using AI, which is amazing because
- 59:52this is a small org that is doing valuable work
- 59:55and something like double maternity leave can,
- 59:59it's just one of those things that you often
- 1:00:01can often come out of nowhere and all of a sudden
- 1:00:04you lose the ability to kind of keep having that
- 1:00:11impact for animals. But he's managed to use AI
- 1:00:13to keep the ship going and things like that,
- 1:00:17which is just really amazing to hear. And another
- 1:00:22thing as well on the front of we've had Another
- 1:00:27organization that I do work with, Greener by
- 1:00:28default, they help caterers to shift to more
- 1:00:32plant forward diets. And so I've been building
- 1:00:36out the data analytics side of their org because
- 1:00:44I'm a data scientist by training. And so I've
- 1:00:48built a system that kind of mostly automates
- 1:00:51all of the data analysis that they need as an
- 1:00:53organization. And I estimate that using AI coding
- 1:00:58assistants, like programming assistants, I have
- 1:01:02been three to four times more productive this
- 1:01:05year than I was pre these assistants, right?
- 1:01:08Like in terms of the amount of code I have shipped,
- 1:01:10the amount of stuff I've delivered, like three
- 1:01:12to four times, right? It's as if, you know, it's
- 1:01:16as if there are several of me in the organization.
- 1:01:19These are the potential improvements where...
- 1:01:22we're talking about here. And because it's such
- 1:01:25a general purpose technology, it can help you
- 1:01:29write donor reports, it can help you research
- 1:01:34for campaigns, it can help you with your HR policies.
- 1:01:40It's such a general purpose tool that it means
- 1:01:43that there is almost every role in the whole
- 1:01:47movement. can make use of AI and can see those
- 1:01:51benefits. It's not just for techies. It's not
- 1:01:55just for young people or people in the West.
- 1:01:57It is really for all of us. And kind of to that
- 1:02:02point, I think it's just about experimenting,
- 1:02:05really. The more you experiment with it, the
- 1:02:07more you're likely to find these benefits. So
- 1:02:11I think if we constantly experiment with these
- 1:02:14tools and we share them, We share the results
- 1:02:18with each other. We share the wins. We share
- 1:02:20the the learnings. Then I think as a movement,
- 1:02:23we can really catch this wave and and just and
- 1:02:28and really kind of use it to have like a huge
- 1:02:32like a much, much bigger impact for for animals.
- 1:02:37I love that. I love hearing those stories. Thank
- 1:02:42you for sharing. And yes, please, people follow
- 1:02:45the footsteps. become more efficient in your
- 1:02:49advocacy work by using AI. I also use AI. You
- 1:02:53know, I'm fundraising consultant. I tell my clients,
- 1:02:57you know, your money is precious. I want to use
- 1:03:00it to be more productive and do more with what
- 1:03:03you've hired me to do. So yes, it is like I said
- 1:03:08in the beginning, you know, it is a necessity
- 1:03:11and people need to start learning to use it.
- 1:03:16And yeah, thank you. Thank you so much, Richie.
- 1:03:21And thank you. Thank you for your advocacy work.
- 1:03:24Thank you for having taken the time to answer
- 1:03:25my questions. This has been very insightful conversations
- 1:03:31I've learned a lot and I'm sure listeners too.
- 1:03:34So we're all very grateful. Thanks. Yeah. Thanks
- 1:03:38for having me. And to all the listeners in the
- 1:03:42new year, keep a lookout for... Thank you everyone
- 1:04:04for listening. I kindly invite you to share this
- 1:04:07podcast with the vegans you know. Let's encourage
- 1:04:10more people to take action. Again, thank you
- 1:04:14so much for caring, and I will see you next Tuesday
- 1:04:17for a new episode.