AI for Animal Activists with Tom Conger

Jasmin Singer: Yeah, we're super excited to chat with you. You are here to talk to us about among other things, my favorite topic, which is AI as it relates to activism.

And as Mariann said before we hit record, I do not exist as a human being anymore. I am obsessed with my AI and so I will be taking notes. This is our Flock Friday, so welcome to our Flock. To those of you who are in the live studio audience, we are so happy you're here. And the Q& A afterwards that Tom will hopefully, unless we offend him terribly, stick around for will be your bonus content this week.

And, anyway, Tom, let's start with a bit of an overview about AI and what it is and what are its implications for animals, good and bad.

We're looking for the 50, 000 foot high view here, and then we'll start to get closer and closer.

Tom Conger: Let's start at that high level of abstraction. What is AI? AI is smart machines, smart systems that are imitating human behavior and making decisions and learning, and creating content just like a human would. We don't know fully what the implications are. There can be some very wonderful implications where AI might help us learn about the vocalization of animals.

And so cows might be able to advocate for themselves to some extent, or at least ask for mercy. It could be terrible where we learn just how cruel we could be before affecting productivity of a factory farm. So one reason why I sort of step out of my comfort zone is because I think AI could be tremendously influential and want us to leverage that as much as we can so that it's helping animals as much as we can.

Mariann Sullivan: Yeah. And anyway, it's here, so we better use it as best we can for animals and try to prevent the worst abuses. Of course, we're not really here to discuss the whole big question of artificial intelligence. But we want to focus on what I've been told is called generative AI and specifically on chat GPT as an activist tool.

And perhaps there are some other programs like it, you know, I don't know anything. So, we're starting at the beginning here, but we'd like to talk about how this is going to change what people do, specifically our listeners, grassroots activists, or also some organizational uses, I guess. And how animal advocates are going to be able to use it.

And so going down from the 50, 000 foot view to maybe the 10, 000 foot view, can you just talk about what generative AI is and specifically chat GPT?

Tom Conger: Sure. I'd be glad to do that. So at the 10,000 foot level, generative AI is a subset of AI as a whole. There's different types. Generative AI, as the name implies, generates content. It could be text, it could be images, it could be video. The format could be poetry or a blog post, and so that's why it's so exciting because it's really allowing us as end users to be able to do things that we couldn't do in the past. I like to think that I could probably create lyrics for songs. That's always been a fantasy of mine to one day publish a song, but I have no musical talent whatsoever. But if I relied on ChatGPT or some other service, there's a possibility that AI could create music for my lyrics, and I could publish a song.

So I really think it's going to be amplifying a lot of our talents, or not so talents that we have.

Mariann Sullivan: Jasmin is now getting off the interview and composing music.

Jasmin Singer: Actually, I was going to say, Mariann, we better take out the question where we ask Tom to sing us a song because he said he has no musical talent.

Mariann Sullivan: That question was not in here, but you know, I'm open to it.

Jasmin Singer: Well, so for some people, ChatGPT has already become a constant companion. I mean, for me, for sure. And I'm sure that for a lot of people, it's just one more thing that they probably should know more about, they think, but they don't. So, Mariann, you're the one who was encouraging me to do it, and now I've just taken it and run.

Mariann Sullivan: Well, it's not like I don't understand that it's important. It's just like, one more thing that I don't understand at all and that now everybody's doing and, this isn't what we're going to talk about because it's too centered on me particularly, but it is on my mind all the time.

This is the first year I will be teaching. I only teach in the spring semester. The first year I will be teaching where the students are really going to be completely on top of this. And it's just hard to know, you know, I want them to learn how to use it. Some teachers ban it, which I think is ridiculous because kids are going to be living with this for the rest of their lives.

But, as I say, I just feel like I need to learn a lot about it, but so far it just seems like a big chore, but I'm told it's so much fun.

Jasmin Singer: I do think we should get back to that. Let's put a pin in that because I think that's a fascinating discussion. But before we do that, I think people need to understand a little bit more about like, let's go down to, I don't know, a 5, 000 foot high level where as we skydive together in tandem, tell us a little bit more about chat GPT and what it does.

Tom Conger: So, ChatGPT, as the name implies, allows a user to use a chat interface. So that you can interact with the AI, with the technology, by providing prompts. And ChatGPT, based on its learning, its neural networks, provides you with a response. And what has been fascinating, and what's really brought everyone's attention to AI, is simply how amazing it is at generating content that is clear, is compelling, that is contextually accurate.

It just gives you what you ask for. And the more precise you can be with your prompt, the better the response is. And the model is not trained just to help you with legal matters or choose a book or a movie. It helps you across so many different things. It can be really versatile and it has the ability to interact with other tools, other sets of information.

So it's this wonderful foundation to build other things on. One of the things that I like about it is it's just so approachable. It's a chat interface, so you ask a question. I'm going to write an email to my boss asking for some vacation, but I've already used all of the time that I've been entitled to use. Can you help me write an email?

And it will help you write an email. Or you might say, I'm really upset about this, and I want to send this email. This is what I've written. Do you think I should send it? Is there a way that I could improve it? And it does a wonderful job of doing that. One of the things I find a little bit humorous is I've been told my writing is very, well, when they're being diplomatic, they say word efficient.

I think I'm the only one that uses chat GPT. And it actually adds words to what I'm writing to make me sound a little more personable. But I like it for that reason.

Mariann Sullivan: That actually totally works for me. I've never heard that term, but I am also very word efficient.

Jasmin Singer: That's one way of describing it. Yeah. Let me just add one thing in there really quickly. I use this email program called SuperHuman, it's a way to very efficiently go through your emails and get to inbox zero and it has all of these tools and they recently incorporated AI. And so there's two things of note, just since you're talking about email in particular. One is when you open an email, it automatically summarizes the email at the top. There's a little box and two is it'll write your email for you if you just give it a quick, this is what I want. I feel the same way I felt in college when I first got email, or I first got on the internet, I should say in college.

When it was like wait a second, are you saying that I can be in chat rooms and talk about Bette Midler all day long with old gay men? Yes, the answer is yes. So that's pretty much what I did my college years, just FYI.

Mariann Sullivan: Explaining why you now need a program to write things for you.

Jasmin Singer: Yeah, well, I did write two books without AI, but I'm not saying that any future books would have no AI, which does bring up the subject of ethics. Do you feel that there are ethical issues at all in terms of creators using AI for kind of jostling their work? What are your thoughts on that?

Tom Conger: I think that if you use it as a tool, you don't have to disclose it any more than you would disclose that you used a spell checker or Grammarly or Google search. It really depends on the extent to which you are reliant on AI to produce the work product. And I know there's a little bit of gray area, but if you're using it as a thought partner, if it generated the first draft, if it generated an outline for you, if it critiqued it and helped you improve it.

I don't think that you have to disclose that you use it, but I do think that in many instances, it probably serves you to say, I produced this in cooperation with or with the benefit of AI, just because it seems to be the appropriate thing to do. I know this sounds a little bit weird, but, you know, Jasmin, you might be able to relate to this if you've been using AI a lot, is you develop a little bit of an affinity for chat GPT or whatever tool that you're using. And I feel like I would be doing it a disservice to not recognize how it contributed to the work.

Jasmin Singer: And also the idea to call it a thought partner. I love that because as soon as I heard that you've used that term with regard to AI, I was like, that's exactly it. I wrote 90 percent of a sub stack the other day and for the life of me could not finish it. And so I put in, I was like, what are some ideas to finish this and it gave me some ideas and I was like, no, that's awful. And that helped. But then I saw one little idea that I liked and I was like, say more about this. And then that got me going. So it was in many ways a thought partner. And so really, really well said. I digress. Mariann, I know you had another question.

Mariann Sullivan: Yeah, no, I have a lot of questions, but I appreciate what you're saying about thought partner and talking about the way you use it, because I know there are some pitfalls and one of the pitfalls that I can see falling into is really just over relying on it, having it write something, not really edit.

I mean, you have to edit pretty intensely when it writes things for you, doesn't it? I mean, I've had it write things and it just gets it wrong. It's not saying what I want to say. It's saying things that may make sense. So how much would you say, is it a tool for writing and how much is it a tool for like, at best creating a draft, which then you really have to go through and make sure that you want to leave it where it is.

Tom Conger: I actually think that it's writing is quite good if you provide the right prompt and instructions. One of the things that I didn't realize when I first started using it is I could tell it what style that I wanted it to write in, and I spent more than probably a day trying to determine exactly what the Stray Dog Institute house style was. Everything from how we were approaching it, our point of view, the complexity of the sentences. And then I realized in some ways we were writing a lot like The Guardian. So then I came up the shorthand just in the style of The Guardian, write X, Y, and Z, made a tremendous difference.

It also responds to feedback. And I'm always polite, and it seems to be polite to me in response, and I just enjoy that interaction more, so I say, great job, I really like that, but I wish that you could make it sound a little more compelling, or less complicated, or whatever it is, and it will respond to that, so the more that you have this back and forth, the better.

I also find that if I, as I'm writing, you can actually say, critique what you just wrote, and then say, what's the criteria that you used to make that critique? What were you using? Well, apply the critique and that criteria to rewrite it. And then to have it continue to produce text, which is wonderful.

In the same way, I could say, here's my writing. Please suggest ways that I could improve it and it will give me suggestions and then I can go and do it. One of my adult children is in school now, taking a computer programming class, and one thing that I really liked is he didn't ask it just to write the code.

He said, pretend that you're my instructor and I'm submitting code to you and I'm having a problem getting it to work. Don't tell me the answer, but show me the path to get to the answer. And chat GPT responded. It said, here's where you probably need to look a little bit deeper into this. And so for him, it was serving as an instructor, a teacher.

For me, it can be a writing coach. You can also ask it to evaluate it in terms of the audience that you're serving. I am writing this to reach young men and women age 18 to 24. How would this go over with them? And, you know, it might say, this is really off the mark, you're writing to older scientists, these are the things that you could do to improve it, and here's an example of what you can do.

So, it can help you tailor things for a specific audience. One time, and I hope I'm not rambling too much, but...

Mariann Sullivan: No, this is all fascinating.

Tom Conger: I asked it to write a piece for a French language newspaper, and it actually, without me telling to write it in French, it wrote my piece in French, which was absolutely amazing.

Now, I don't speak French, I have two colleagues that do, but I would never send that to the publisher without having someone that knows the language read it and review. In fact, if you talk about missteps, that's probably one of the biggest missteps is not reviewing the content that AI produces before it goes out, whether that's an email or publishing. It gets so tempting because the better the prompts the lower the risk, you could just have things go out, but that's a recipe for disaster. Always be an intermediary between AI and the recipient of the data.

Mariann Sullivan: Well, it does this thing called hallucinate, right? I mean, maybe I'm more afraid of that, but there's this famous story within the legal world. Maybe this is famous everywhere. I don't know. But some lawyer, and this was a while back, maybe six months ago or so. And I know things keep improving and keep improving.

But had chatGPT write a brief and submitted it. It looked great, said what he wanted to say. ChatGPT made up all of the cases out of thin air. Like apparently chatGPT, without sounding too much like I'm talking about it like it's a real person, understood that there are these things that you intersperse that have names in them and the name of a court.

And it just made them up because it didn't know how to find them, I guess. I don't know whether that's even worse for lawyers than for other people, but checking everything would seem to be really a good idea.

Tom Conger: Yeah you need to check to make sure that the information is accurate. And just so you know, I'm not an attorney, but I heard about that example.

Mariann Sullivan: I figure it's probably famous throughout the world.

Tom Conger: It is, and at some point we should probably talk about the implications to law, because there's certainly some, and it's quite interesting. I think the legal profession is really ripe for disruption around AI, and I think the changes are going to be quite significant.

But in addition to misinformation, I think you have to be careful to check for bias. It's probably most evident when you look for images. So, create an image of a doctor talking to a patient. Well, because AI, including ChatGPT, is learning from this vast amount of information that's out there, it's going to replicate many of the biases that are out there.

So, the doctor in all probability is going to be a white man. So you have to be careful about that. Those biases are going to show up in written language as well. So you'll notice, for example, it might be producing an article about a cow and know the gender and refer her as it.

But there are things that you can do to control for that by describing the race or ethnicity or whatever profile that you want, or to say, I want to have something that is ethnically diverse, or whatever category it is that you're looking for.

Jasmin Singer: I do want to say that I just opened chat GPT and I wrote, create an image of a doctor talking to a patient and it is a woman talking to a man, but you're absolutely right. And honestly, what you're talking about now is I think a little terrifying.

Tom Conger: Well, it is, but one thing that I really like about all the companies, and I know part of it is marketing, but I think also part of it is this real concern that this is very powerful technology and they want to use it in the right way, is they create the AI, but then they provide feedback on what it produces and they provide guardrails.

So one of the reasons that you might have seen that image is because they've recognized that there was bias built in and they're helping to correct for that bias within the data set. So, I think it will be less of a problem over time, but one still has to be careful of it.

Jasmin Singer: Totally. Mariann, go ahead.

Mariann Sullivan: I was really gonna change the subject because I don't want to forget to ask if people who have not yet used chat GPT are getting curious, can you talk a little bit about how you find it how you use it, which couldn't be simpler but also crucially the difference between 3. 5 and 4. 0 and what you would recommend?

Tom Conger: I'd be glad to do that. So, ChatGPT is developed by OpenAI. OpenAI.com is their website. You go and you can sign up. They had so much interest that they're, at least for a few days, not allowing people to actually purchase their subscription level to chat GPT 4.

So if you go and it's not available, hang in there. Eventually they will make it available again. The difference between 3. 5 and 4. 0 is substantial. And one of the things that I've always been concerned about is people go to OpenAI. com. They sign up for 3. 5 for free. They test it out and it's like, hmm, okay, I don't really understand why everyone's so excited about this.

The difference between 3. 5 and 4. 0 for almost everyone is well worth, the money of 20 a month. It produces more accurate responses. It can understand better the prompts and where one falls short on the actual prompt, it can kind of help to compensate for that. It can interface directly with plugins that can allow you to access other information and other sites.

You can upload reports to it. It can help you with prompts that you can then give to Dolly, which is an image generator. And so you can access that image generator directly from ChatGPT. So, if you want to figure out if this whole ChatGPT thing is for you or not, spend the 20, access 4. 0, try it for a month, and then make your decision. Don't base it on 3. 5. The difference between the two are quite substantial.

Jasmin Singer: Yeah, I totally agree with that. It's funny when you were talking about how we need to know how to guide it. I think that's a skill and I think it's a skill that a lot of people could actually stand to get a little better at. I will tell you that in two different jobs where I had male bosses, I was called bossy.

Which, I'm laughing because it's like, dude, don't say that to a woman, A, and B, I'm actually not bossy. I just have confidence about working with people and I know, generally speaking, how to work with someone in order to get the product out that I want to.

And I think that that has been a large part of why I love working with it so much, because I think I, I know... basically it's my little bitch, is what I'm saying. But, that aside, how do you use it, Tom?

Mariann Sullivan: It'd pretty funny if it responded to one of your prompts by saying don't be so bossy

Jasmin Singer: It'd be hilarious. but yeah, go ahead, Tom.

Tom Conger: Well, one thing that's interesting is within ChatGPT, you can actually provide what are called meta prompts to provide a little bit of personality to ChatGPT and how it responds to you. So if you wanted it to call you bossy or boss, you could provide those instructions. If you wanted it to be nice because you're thin skinned, it would be nice.

If you wanted it to be direct, it can be direct. So you can kind of describe how you want it to interact with you. But I agree if you're used to providing detailed instructions, that really helps. And when people say they're not satisfied with the responses that they're getting, invariably it's because the prompts are not as good as they could be.

I think for a year or two, there might even be a profession prompt engineer, which is a term that I've heard people use, because they're really skilled at saying, This is the prompt that you should use and write to get specific kinds of output. I think over time, and we're already seeing this, ChatGPT is saying, okay, here are the five words that they put in as a prompt, but what they really meant to say was x, y, and z, and in essence create a larger prompt with more detail that you can then edit.

I've even asked ChatGPT, what did you think of my prompt? How could I have improved my prompt? And it will give me information, and usually it's providing more context around audience and purpose, and who am I? Am I an animal advocate? Am I a scientist? Am I an attorney? Am I a journalist? All of that information serves as a map so that when AI is going out and looking at its data, it knows where to go and find the information and the words and the concepts that could be most relevant to the prompt.

Jasmin Singer: Yeah, so true. I'm curious about how you use it. You've talked a lot in generalities, but tell me a little bit about a day in the life of Tom as it relates to AI. And then, we're going to pivot and ask you about how other people, especially those who care about changing the world for animals can use it.

But let's talk about you first.

Tom Conger: Okay, so, I have quite an affinity for ChatGPT. I spent I don't know, two weeks probably ignoring more emails than I would like to admit because I was just fascinated by it. You know, Jasmin, you talked about remembering the first time you got an email and what that felt like. For me, it was burning my first DVD, how powerful I felt and cool that I could do this and what a step forward it was. And now that sounds just so old school.

I was in a driverless taxi in San Francisco earlier this year and thought, Wow, this is just amazing. And it even did things like, to switch lanes, it kind of did a little jump to get the other traffic to stop, which I thought was really clever.

The feeling that I got in working with AI was like 10 times as thrilling as the driverless taxi. So I always keep AI open on my desk because if I don't, I forget about it and I just stick to the same routines. I deliberately ask myself and ask ChatGPT, this is the task that I'm working on, how can you help me?

And I'm surprised at all the ways that it suggests that it can help me. Sometimes I'll simply take something that I'm working on, I'll paste it in and just see what ChatGPT says. And it actually does this wonderful job of... sometimes it makes it up, but sometimes it infers. Tom, I think this is what you're working on. You're working on a mission statement for a non profit organization. I like it. It looks good.

And then I might say, can you improve it? And all of a sudden, so it's a way of kind of always pulling me back to using AI as a thought partner. So keep it open at the desk, even if you think it can't be helpful.

Stick some content in there, see what you get, ask questions, ask for advice, that's number one. Number two is don't underestimate it's power, because with the right prompts it can do simply some amazing worK. So in a previous life, I was a management consultant. I specialized in foresight strategy and innovation and did pretty well to write scenarios for organizations that would talk about critical uncertainties and how the future might unfold.

So on day two of using ChatGPT, I said, let's see if we can push the limits of ChatGPT and what it can do. So I said, let's think about the global food system. Let's look at the next 25 years. What are the critical uncertainties about how the future might unfold? What are those things that can have a tremendous amount of impact, but there's lots of uncertainty. Create a list for me.

It created a list. It was pretty good. I added some. I said, what did you think about those? It gave me some feedback. So after about an hour or so, we had a really good list of critical uncertainties. And then I said, let's talk about the food system and the way that we want to describe it, particularly those things that might be variables in the future.

What are those aspects of the system? Created a list. We did some back and forth. And then I said, okay, let's look at the impact. So create a cross impact matrix for me of how those critical uncertainties and those aspects of the future interact. And I'm like, Oh, wow, it produced a table for me. I didn't know that ChatGPT could produce a table. And then I said, let's take it to the next level. Let's pick the two most important critical uncertainties. So I said, which are the two most important? It gave me two. I said, why? I said, Oh, I like that criteria, but let's work on that a little more. So we've revised the criteria and then together we picked two and then we picked sort of a continuum for each of them. Oversimplified, but let's say high, low for both of the critical uncertainties. And I said, let's pretend that it's 25 years in the future. These are your two critical uncertainties. Let's create a two by two matrix. And I want you to write a narrative for each of those four scenarios about how we got from here to there. So within a day I had produced what I thought were really compelling and coherent scenarios that showed pathways to these very different futures.

And I had a sigh of relief. I'm glad that I'm not in that business anymore because it really got me to a point that would have taken maybe two months in the past, to get to that stage. So always, always test it. Another example, we created a rubric, if you will. We wanted to look at how we were spending our time at Stray Dog Institute and where our philanthropic dollars were going.

And so we created a matrix of levers and focus areas. And we wanted to know how our philanthropic investments were spread out across those areas. And then we realized, well, some organizations are going to work across areas. So I actually submitted to GPT, the matrix, and I said, here's one of our nonprofits.

Here's their website, and ChatGPT now has access to the internet. It used to be a limitation that it didn't. And I said, go out and learn about this organization, and then come back and just take your best guess at how they allocate their time across the cells within my matrix. Government, public, business, movement building, you know, were they using levers like legal, you know, whatever it might be.

And it came back and it gave me a distribution of how they spent their time. And I'm like, well, that's pretty good. You know, I'm not sure if I buy it. Then I said, just limit it to five cells. You can't use any more than five. And then it allocated 100 percent across those five.

And it gave me something that was quite meaningful. So, it's very easy to then say, wow, I could go out and I could say, look at a hundred different grantees that we've supported and map them and their allocation of resources across these areas to come up with a map for us. So again, that's a very complicated way of using it.

But, it just gives you some sense of the breadth, if you will, of the things that. that you can do.

Mariann Sullivan: Wow. That's totally amazing. One thing I'm wondering, you know, what I use to find out things on the internet and what everybody has used is Google, and they're very, very different things, but they're not completely different things. You sort of use them for the same thing.

I mean, specifically for research, not the kind of internal work you're talking about, but like the first thing you were talking about, looking into food systems. And making it perhaps a simpler issue, but when you go into Google, it gives you all these different sources and you can look at them and read them and it's much harder because you have to read everything separately, but you do have the ability to evaluate what you think is a good source and what you think is not a reliable source, an industry influenced source or whatever.

I'm just wondering, is that something you can take into account with ChatGPT? And is it also something that could be gamed? Like if the industry really wanted to bury a lot of information about what they're doing to pigs or something that they could really lean into... cause we're a tiny movement and it wouldn't be that hard to flood information out there to overcome whatever things that we're putting out.

I hope that question was clear. But it has a lot of combinations of how do you know whether to trust it with the research it's doing?

Tom Conger: You can't always trust it. You mentioned hallucinations, so it can hallucinate. Again, I know we're putting these human characteristics on it, but I think sometimes it really wants to accomplish the tasks that you've assigned to it, and so it does its best, and if that means making up cases, it's going to make up cases in order to make you happy.

Chat GPT, particularly if you use it within Bing, will provide you with a source. So, you can actually say, tell me about AI and how is it related to machine learning or deep learning. It'll give you a response, it'll provide a footnote, and you can actually click on that footnote and check out the source. It still does not compare, in my mind, to what you get from Google in terms of real time data, the most recent information. It's certainly a step in the right direction to provide a footnote, but I think it's going to be a while before ChatGPT replaces, or could replace, what Google is now offering.

I think the tools go hand in hand because if I didn't know anything about the food system and I wanted to understand it and do some research and I went to Google, I'm going to have to sort through a lot of information before I have this sense of, of a map. Whereas if I go to chat GPT, I could say, what are the top 20 factors that are part of the food system and how do they relate to each other?

And it might say, boy, that's a big question. I'm going to give you the 20 factors first, and then we can sort of dive into it. And so in 30 minutes, an hour using chat GPT, I understand the map of what the food system looks like. I understand how the factors might relate to each other. And then that guides the work that I would do on Google.

So I think for the moment they go hand in hand. A lot of it depends on the task. If I want to understand something that's historical, that is a concept, ChatGPT is wonderful. If I want something that is certainly more timely, might prioritize some information over others and do so in a more reliable way, I would go to Google.

One thing that I don't know is when chat GPT provides a resource, sometimes I think that it, and I have no idea because again, I'm just an enthusiastic end user, I think it determines what it wants to say, for the most part, and then says, let me see if I can find a citation that will support this point of view.

Mariann Sullivan: That sounds exactly like being a lawyer.

Tom Conger: So sometimes I wonder, is it really picking the best source? Mariann, one of the things that I really like is you can actually ask it about the sources that it's using and and say, do you think that this was influenced by animal agriculture industry or the meat industry?

And sometimes it can make some connections. You know, this is a land grant university and there's a tenured professorship that is sponsored by a meat company. So, again, I think there's some insights.

Mariann Sullivan: Unbelievable.

Tom Conger: I think both sides of the animal agriculture fight are thinking about what information that can they put out there that can help influence how... yeah. So if we all started using gender pronouns when we talked about animals, when we knew about the gender of the animal, I'm not recommending that one way or the other, I'm just using it as an example.

Would that help chat GPT and other AI begin to talk in terms that were less commodifying and not saying it, but really recognizing their personality and their sentience. And so, yeah, we could all try to flood. There's this common expression about, will AI start eating its own tail because AI is going to be flooding the market with information and is AI going to be relying on that information to create new content?

So, there are already podcasts right now that's comprised of only AI interviewing other AI. So, just to practice a little bit before I had this conversation with both of you, I used my phone, and one thing that's nice about ChatGPT, it's multimodal now, so you can take a picture and ask it, what is this of? What could I do with it?

So, those are yellow onions, you use it in cooking, this is how you might use it. You can give it pictures. You could show it what you have in your pantry and it'll say, here's a vegan recipe of the things that you can do with that. I talk to GPT now.

So you can say, instead of typing, you can just speak the prompt and then it can respond to you in a voice that you choOse. And one thing that's interesting to me is I tend to talk with lots of pauses and ChatGPT thinks I'm done. So I do feel a little bit rushed when I'm talking, but it's really great because you can have these conversations.

You could say, could you tell me more? It even suggests prompts of things that you might ask. So I give a description of what AI is, and then I can say, well, how did I do? So anyway, that's a, a little bit of a departure, but, it is a real concern about how AI is going to respond to a flood of AI generated information. And, you know, it's supposed to be learning about how humans interact, but if instead it's learning how AI interacts with other AI that's going to skew the system. There are even influencers now with hundreds of thousands, I don't know if any of them have millions yet, that are AI influencers.

They're not real people, they're just AI generated images and AI generated content. So, obviously, I thought, Well, I wonder if we could create millions of vegan influencers, that 95 percent of their content has nothing to do with veganism. It has to do with any number of things, plants or woodworking or stock investments.

And then they just happen to mention every once in a while that they're vegans and why they're vegans, right? Could we do that?

Jasmin Singer: I love that, that's so good.

Tom Conger: Yeah, so anyway, you can tell I get a little excited about the possibilities.

Jasmin Singer: Oh, you're such a nerd.

Tom Conger: I am. Thank you!

Mariann Sullivan: I find it so exciting and so terrifying at the same time. And so perplexing, but you're helping.

Jasmin Singer: Yeah, the comments, by the way, are so funny. Let me just read a couple of them even though I said I wouldn't. Vicki said, I think I need to look into AI more. I got so annoyed that it was being used for the arts that I sort of started just hating it as an idea. And then Jen said, yeah, I didn't put a lot of thought into it because I didn't know the depth of it.

And then, you talked about the recipe thing and Vicki said, shut up with the recipe thing. I need it just for that. And then Sandy's, this is my favorite. Sandy said, Vicki and Jen, same here. I thought all it did was help people cheat! That's so funny.

Mariann Sullivan: I do want to talk. Oh, I'm sorry.

Jasmin Singer: No, we could save this for the Q and A.

I'm just getting ahead of myself. I did want to ask you, Tom, how our listeners and people who want to change the world for animals can get involved on a day to day basis. You mentioned some of the ways you do, but what specific ways do you think that activists or advocates or just your general human being who gives a crap can use chatGPT?

Tom Conger: So, two things come to mind. One is being able to do something without having to make a huge commitment of time. I don't know how many requests I get, you know, send a letter to your congressperson, send a letter to an organization, I'm like, I want to do it, I'm going to set it aside. And, I just don't always get back to it.

Mariann Sullivan: Yeah.

Tom Conger: Now, I can put it into chat GPT and I can say, create a letter for me that I can send. So, in two minutes I have something. And it's really easy to do because you can simply say, Here's the request that I received to write a letter. Write a letter for me. And one of the things that I'd like to emphasize is that, their headquarters are in Northern Virginia, and I live in Northern Virginia.

And it could add that kind of content and appeal.

The other thing that'll be really interesting, and I'm sure it's just a matter of time, but there are platforms that are out there that allow you to write your congressperson or write a letter, and it says, here's the boilerplate, right? But, you have to take the time to write it differently.

Well, I think in the future, it's going to say, Hey, Tom, I recognize that you're back at this platform again. Thank you for sharing your demographics. I actually looked at how you edited our letter last time, so I know what your writing style is. I checked out your LinkedIn profile and you know years ago you could look at someone's LinkedIn profile and have systems that would take a guess at their Myers Briggs type indicator so you'd know how to pitch or sell to that person.

Well now it could use that to understand what your style is and who you are and compose a letter for you. So one way that you could use it is to do all of those things that you know you want to do but don't have time to do. You can now do it efficiently. So you just don't have the excuse of time anymore.

So that's one thing. The other thing that I think makes a huge difference, if you're like me, there are some areas where I feel really confident. So I feel like given enough time, I could write a decent blog post or letter to the editor. Writing is a strength. It's a lot different if I have to interact in real time with people.

And I can use ChatGPT as a way of building my confidence. So you can actually ask ChatGPT, be a reporter that's asking me about Stray Dog Institute and my views on the food system, ask follow on questions, listen to my response and dig deeper, critique what I'm saying. And so the first time I used ChatGPT in that way, I was like, boy, it is so different to respond to these kinds of interactions in real time as opposed to in writing.

But after doing it a few times, I'm like, okay, I kind of get the hang of it. You can even ask for some advice. If you're volunteering or you're going to be at a protest, you can talk about the different types of interactions and the things that might happen. So I, you know, I think in the future, organizations are going to prepare their advocates for different types of interactions with the public in different settings and ChatGPT or other tool is going to help build their confidence so that they can feel more comfortable about calling their representative or going to their office and stating a point of view.

So saving time and building confidence would be two. And then I've mentioned being a thought partner. You know, obviously, if you're doing campaigns, you could say, help me write posts for social media on this platform. Give me 25 ideas. You could say, I like numbers 3, 7, and 8. Let's build off of those. What are your favorites?

I always like asking, what are your favorites? Why are those your favorites? That gives me the criteria, and then I could use that criteria to go back. I could say, I really want to appeal to men who are 50 years of age and older, or women who are 18 to 30, whatever it might be, and it will take that into consideration when it's producing the content.

You can use it for campaign materials, you could use it to design campaigns. The more sophisticated software in the future, you're going to be able to access your database and profiles of folks and what they've done, and it'll actually say, here are the 10 people that you should reach out to to help you with this specific campaign, or where you can go for funding, or whatever it might be.

And then, if you don't mind, I, I know I said two, and then I'm already on my fourth, but one thing that I have found really helpful is having a test campaign ideas. Because I've had ideas for campaigns that I've carried around for years that I'd love to do one day.

And I thought they were good, and so I could never set them aside. But I could work with ChatGPT now to really dive into a campaign and say, This is what I'm trying to do. This is how I'm trying to affect change. This is the campaign that I've designed. What do you think about it? And it will give me some ideas, and then I can ask it for alternatives. What are 10 other ways that I might accomplish the same thing? And then it gives me better ideas than I had before, and I can set that idea aside and create this new space in my head to tackle other things. And so, that's another way that you could find it useful is Asking it, is this really a good idea?

Can I improve it? Should I be doing something different that could accomplish the same objectives?

Jasmin Singer: Yeah, same. Just to add to that, I have pasted articles in it that I've read, and I have said, what would a critique be of this article as a vegan? I kind of want to see if it matches what's in my head, and sometimes I'll find new ways of thinking, too. So, for some of you, like Sandy, who's commenting, I feel like that would be really good for you, because you're very involved as an activist in that particular way, you know, and just kind of getting new ideas going and things like that.

I also wonder if moving forward, we're going to need better editing skills. I'll be curious during the Q and A, some of our flock members here are writers. And I'm curious what you would think of that because I have experience editing. And so maybe that is also working in my favor.

Maybe it's another reason why I like this, but you do have to be able to go into something and change it, and you have to have your editing brain on, too. One thing that was super creepy for me is that I had it write something once, in the style of Jasmin Singer, just to see what would happen, and it worked.

It was like, slightly hyperbolic, and using some personal experiences, and it's sort of fascinating. I know we can't keep you on for that much longer because I want to get to the Q& A, but there is a question from a Flock member that I would like to ask here for the full episode, one of our Flock members says, there's been talk about ethics regarding disclosure around whether you've used AI in creating something and around awareness of potential biases.

But what about the ethical questions regarding underlying copyright issues in what has been used as the training material for the AI tools?

Tom Conger: That's something that will have to be resolved in the court system. I'm really torn because if I had made an investment in writing poetry or music, or books and had a style that was me, it's Tom Conger, it's my point of view, and someone is copying that to produce content that sounds like me, it's really hard to accept that.

At the same time, I recognize that anyone could adopt that style, it just makes it really easy that AI is doing it. But if I wanted to copy Jasmin's style, I could do that. I could, at least I could try to do that, and be somewhat successful. So again, I'm not sure how it's going to play out. I don't know if there'll be some compromise in which content creators are compensated in some way when it is used. But my guess is, if I had to take a guess, that anything that's available in the public domain or available for purchase, is going to be fair game and that, just like a person could access that and use that knowledge and that style to create content, that AI is going to be able to do that as well.

Particularly as we get to a point where we start to think of AI as doing things that are very human like. Nothing is going to keep a person from copying the Jasmin Singer style, if they want to. Now, if they say Jasmin produced it, that's totally different, right? So there'd have to be some disclosure.

But, I don't know how it's going to play out. I don't know if they're going to avoid court by having some modest compensation for folks, but it's really interesting and I don't know how it's going to play out. But the one thing I will say is another thing that could end up happening, and you run into this with people reverse engineering product. So if you have one team that does the reverse engineering and that same team produces the content, that's a problem. But if you have one team that takes everything apart and says, these are all the design principles, this is the technology, and then you break it down into its component parts, and then you give that to another team to reassemble, then it may be fair game.

And so I think what will end up happening perhaps is they're not going to say, Copy Jasmin, it's going to say, describe Jasmin's writing, personality, style, point of view, and create a description of that.

Jasmin Singer: Bossy!

Tom Conger: And then you take that bossy, and then you take that, and you plug it into another system and say, here's the style that I want you to emulate.

We've never talked about Jasmin, but it's clearly a description of Jasmin. And then it produces the content. So legally, that's how it might work in the future. I have no idea, but it's fun to speculate.

Mariann Sullivan: Now, I don't know how you could have any idea, because I just feel like the entire world is changing. It's impossible to predict how this is all going to play out. It's just such a huge change in the way content, for lack of a better word, it's not a very attractive word for everything that we do, everything that humans do. It's all going to shift.

It's just mind blowing, it really is.

Jasmin Singer: So before you go, and Tom, honestly, I'm kicking myself because we have so many questions for you, and I just want to keep you on for the next five hours, but we can't.

Tom Conger: I'm having fun, so keep me as long as you want.

Jasmin Singer: I'm giddy. I don't know if you can tell, but I am literally so excited about all of this, and I know people are scared, and I'm scared, too, to some extent, but I'm more excited about the opportunities.

And I love what you said about the possibility of AI influencers and, like, making them vegan and I think that would be an amazing use of funding, to make sure that that kind of thing is funded. I thought it was absolutely genius when Sentient Media, for example, made such a big focus of their work SEO and making sure that these articles that they were placing in mainstream media were super Google able and kind of infiltrating the normal world, whatever that means.

So love that idea so much. I hope you come back again, but before you go, and I mean, this is ridiculous to ask this as like a side note question because we could 1000 percent do an entire interview about this, but tell us about Stray Dog, what it is and what it does.

Tom Conger: Stray Dog Institute is a private operating foundation that was founded and is funded by Chuck and Jennifer Lau. They, as the name implies, came to the animal protection, animal advocacy space from companion animals, but also realized just how horrific the animal ag system was, particularly around factory farms.

As they started to become more concerned about farmed animals, they realized there were a lot of other food system issues as well. So a lot of overlaps with other oppressions, whether it's workers or the impact on the environment of rural communities. And so we center farmed animals in all of the work that we do, but we have an awareness of the food system.

And as we advocate for reducing the use of animals in the food system or replacing them or reforming their use, we always have in mind how it impacts the rest of the food system. We certainly don't want to cause harm. We also want to look for co benefits. So, if you're fighting factory farms, it's great for animals, but it's also good for farmers.

It's also good for rural economies, for the health of the public, so do no harm, look for co benefits, and then there are aspects of our work as well that get into food system issues that aren't immediately tied to animals, but they are important. I mean, they may not be our top of our philanthropic priorities, but for example, we did several years of work on farm transformation.

If we're concerned about animals, we should be concerned about farmers. So how do we transition them in a just way? Just like we would try to transition coal miners to find other jobs, it's the same with farmers. We realize there's a lot of racial justice issues within the food system, so some of the funding is going to improve racial justice and social justice issues within the food system.

Again, farmed animals are very centered in our work, but we have an awareness of the food system, and we do spend some time and money in some of these other areas. So that's what we do.

Jasmin Singer: Amazing. What are you most excited about that you're working on?

Tom Conger: I would say that we have a better understanding now of how we can support the movement in ways that we might not have fully understood in the past. So we've done a pretty big transition philanthropically. And it's all relative, right? So, we have quite a few fewer zeros in our checks than some of the larger funders in the space.

But for us, we would write 20 large checks a year. And now, we're writing a hundred- 150 smaller checks, which allows us to support the movement as a whole. And so, I really like supporting the movement as a whole. We're supporting a wider variety of interventions. We're supporting smaller organizations.

We're making an effort to identify and support BIPOC led organizations. We're moving part of our attention outside of the U. S., which I think is really important. We've been adopting the tenets of trust based philanthropy and how we relate to non profits. I think in the past it tended to be, hey, we'd like to work with you as a non profit. Here's a big check and let's talk about exactly what you're going to do with that money. So now it's more about let's work together as a thought partner where we have something to offer, but they have a lot to offer as well, and we trust each other. We're in long term relationship with the nonprofits to figure out the best path forward.

So this. Supporting the movement as a whole, supporting the supporters of the movement, Sentient Media, Vegan Hacktivists, are examples...

Jasmin Singer: And Our Hen House, I just want to say you support Our Hen House and we love you for that because not every foundation sees the value of media and we just appreciate all that you do, but not only that, but the fact that you recognize that we need a diverse, multifaceted approach to change-making.

Tom Conger: Exactly. Like you said, there's probably a whole show that we could talk about our theory of change and how we approach things. But yeah, I'm just most excited about how we're relating to the movement and how we can be more supportive. And being in relationship with so many different organizations and how much we could learn from them.

It's just been wonderful.

Mariann Sullivan: And you mentioned that in the time that you've been doing this, you've been in a learning process about the movement, but don't you also feel that the movement has just expanded dramatically, I don't know, the past 5, 10 years? Or the past two years! Like there are just so many more organizations internationally, doing just amazing policy work.

The whole movement just seems so much more sophisticated than it was 10 years ago.

Tom Conger: Yeah, it is becoming more sophisticated and world class in how they approach things and the platforms that they're using and the level of sophistication. The narrative work has been improving and people are testing messages now. A lot of groups, you know, Pax Fauna, Animal Think Tank, are really testing different messages and what works and doesn't work.

There are groups that are helping vegans reach out to others in a way that's not offensive and one of the things I did with ChatGPT early on was I said, help me understand why people still eat meat, just remind me.

But it's just this reminder of the motivations that people have. It's the tradition or the associations with masculinity or their identity or whatever it might be.

Anyway, I just feel like there are so many success stories in the movement. I know there are setbacks, but there are so many success stories that I feel like we might actually... you know, I'm older than a lot of young advocates, and so I'm not sure...

Mariann Sullivan: You're not older than everybody, Tom.

Tom Conger: Not older than everybody, fair, fair enough.

But I feel like there's, and I know I'm an optimist, but there's just so much, success that we're seeing, particularly with young generations and their embrace of plant-based diets. And even if they're not doing it for animal welfare issues, even if they're doing it for health or for the environment, it's just very encouraging, very motivating. And I just feel like it's a really exciting time to be part of the movement, and I think that AI is actually going to help us.

I know that some folks will talk about AI and how the meat industry or animal ag industry will use that, but I think they're already well resourced and I think it's the people who are under resourced where ChatGPT and other AI tools can help amplify our work.

It's gonna be much more advantageous to us than it will be to the big guys. That's what I think.

Mariann Sullivan: That makes so much sense and really is a place for hope. I mean, this tool is much more valuable... I mean, they have all the money in the world to spend on their nonsense. So it's not gonna change their game a lot, but it's gonna change our

game completely.

Tom Conger: I think so.

Jasmin Singer: So tell us where people can find out more about Stray Dog.

Tom Conger: Straydoginstitute.org. I do hope people take the time to look through our blog posts, if you wanna understand our point of view and how we approach the food system and how we try to elevate the voices of our non-profit partners.

Go there. You can really learn a lot from our blog post. And it wouldn't surprise me if next year you could go to our website and interact with a chat bot and actually ask questions about our point of view, and it will pull information from our blog post to help answer those kinds of questions.

Jasmin Singer: Fantastic. I am so grateful to you for all that you're doing to change the world for animals and for really giving people some incredible ideas for how they can get involved with using AI to help change the world for animals. So hang out because we're going to do a Q& A with our flock, but thank you so much for joining us today.

We really appreciate you, Tom.

Tom Conger: Well, thank you for having me, and Mariann and Jasmin, thank you so much for what you do for the movement. It not only impacts media and what others learn about our movement, but I think it really creates a sense of community, and as a social movement, that sense of community is really, really important.

So, I'm really grateful that you invited me to participate and contribute to the community. I'm so grateful for the flock and everything that they're doing. Thank you.