Asking a Squirrel to Explain Politics

Paul Ford: Hi, I’m Paul Ford.

Rich Ziade: And I’m Rich Ziade.

Paul: And you are listening to Reqless, the podcast about how artificial intelligence is changing the whole big world of software.

[intro music]

Paul: All right, Rich.

Rich: Yes, Paul?

Paul: We should just, let’s do the sponsorship right up in front. I am the co-founder of a software company called Aboard, and I’m joined by my other co-founder, Rich Ziade. We also happen to be veteran podcast hosts. What’s Aboard?

Rich: Aboard uses AI to build custom software really fast for you. And when we say fast, that means it’s cheaper. Have you ever heard the phrase “time is money”?

Paul: Boy, I have.

Rich: Nobody knows that phrase better than consultants, product managers, business analysts, UX designers, UI designers, UXD designers, IX designers—

Paul: I mean—

Rich: —content strategists—

Paul: I mean, I think you’ve made that point. Look, the whole point of, the reason this thing is called Reqless is you can skip the whole requirements phase, which, you know, a million consultants just burst into flames when I said that. But the reality is you can get software—AI, man, after 6 million years building software, AI is really good at slapping software onto the screen faster than you could ever do it before. That’s why we’re called Reqless. You can skip the whole requirements phase. You can get stuff up there. And it is cool.

Rich: A quick shout-out to the consultants, though, Paul. It sounds like we’re throwing you in the river. There’s plenty for you to do. It’s just different.

Paul: Wow. This is a hell of an ad. This is really good. Like, we really nailed it down.

Rich: No, no. [laughing] Hold on—

Paul: Okay.

Rich: For example, we can teach you how to make an omelet.

Paul: Mmm. That’s true. That’s true. It’s time for consultants to reskill. That’s what I can’t wait. They’ve been telling everyone else they have to reskill for 30 years now. [laughter] I cannot wait to give a McKinsey consultant a hammer and say, “Let’s go.”

Rich: We’ve talked about this. Let’s actually touch on this real quick.

Paul: Okay.

Rich: Tech has always required people in the tech space to sort of adapt and, level up as innovations happen.

Paul: Mmm hmm.

Rich: There was a day where you needed someone who is an expert in writing batch scripts, and that’s different now. And DevOps is a ten-year-old profession, not a hundred-year-old profession. So it always shifts around. I think it’s a really interesting time. Check out Aboard at aboard.com and also, we’d love to talk. Reach out to us.

Paul: That’s right. Be in touch. Be in touch. All right, so, Rich, I want to throw one of my classic curveballs into the conversation.

Rich: Okay…

Paul: I don’t know if you know this, but there is a presidential election that is going on in America right now.

Rich: It’s the year of our lord 2024. It is August, and yeah, there’s an election coming up in three months in the United States.

Paul: And—

Rich: If you’re listening to this podcast on the Internet Archive 15 years from now.

Paul: Boy, I would love to be 15 years from now, frankly. [laughter] So one of the big subjects, the Republicans are associated with an organization called the Heritage Foundation, and the Heritage Foundation has funded this big thing called Project 2025. It’s kind of much in the news.

Rich: Yeah.

Paul: And it actually took form as a 900-page PDF about how when Trump is reelected, they will enact all of these very, very broad—they’re going to consolidate a lot of power in the executive branch, and they will enact all these very, very broad, sweeping cultural changes that kind of align with very conservative values.

Rich: Okay, fine.

Paul: Okay, well—

Rich: You lost me on 900-page PDF, frankly.

Paul: Well, you lost me on that, too. Unfortunately, the Internet that has ruined my brain. I do not have the ability to read a 900-page PDF from a relatively far-right organization anymore.

Rich: Is the 900-page PDF meant to be read?

Paul: I don’t, you know, people who like politics will do that stuff. It is just—

Rich: Yeah, yeah. Think tanks and positions papers and such.

Paul: I don’t like politics. I just am in a state of panic. And so anyway…

Rich: Okay.

Paul: I was like, how can I understand this thing? And now there’s better resources, but this was about two, three months ago. And I was like, well, let’s see if the bots will help. So I uploaded it to ChatGPT. Normal thing to do.

Rich: Okay. The whole thing, the 900-page PDF.

Paul: 900 page, it’s a lot of words, but whatever, right? So I was like, read this PDF and answer questions about it.

Rich: Okay.

Paul: And I’m going to confess something about myself. So I asked it a few questions. Still kind of boring.

Rich: It’s sort of regurgitating what was in the—

Paul: Yeah, it’s just summarizing the points. So then I said, okay, can you please answer the questions in the form of a really cheerful, optimistic squirrel?

Rich: What?

Paul: Yeah! And I’m going to give you an example, and I’m going to tell you something, I learned more about Project 2025 by talking to the squirrel than I have from any other source in the media.

Rich: Okay, are you going to pitch-shift your voice now to sound like a chipmunk?

Paul: [laughing] No, let’s not do that. Let’s not. Let’s not do gimmicks on this podcast, this is serious. Serious journalism.

Rich: Okay. Do you have some answers you’re going to read out to me, Mister Squirrel?

Paul: Here’s what I said. I said, how would school lunch programs do under Project 2025? Okay? And it said, “All right, let’s nibble through the school lunch programs under Project 2025, [laughter] Squirrel Style.” And then it puts a squirrel emoji and a little what looks like a peanut butter sandwich emoji.

Rich: Okay.

Paul: Okay.

Rich: All right.

Paul: So we’re doing good. Now here, I’m gonna just give you a couple of the answers. There’s six. Gave me six bullet points. One, local nut control. Project 2025 wants local squirrel communities to have more say in how school lunch programs are run. This means local school districts would decide what kind of acorn and berry lunches to serve, rather than following strict federal guidelines.

Rich: Okay…

Paul: All right. Okay. Local food. Two: Private-sector snacks. They support bringing in private companies to help provide managed school lunches. Think of squirrel entrepreneurs setting up nutty snack stations and lunch carts to make meals—

Rich: Oh dear God.

Paul: [laughing] —more tasty and efficient. Two: Healthy nut standards. While they want more local control, Project 2025 still emphasizes healthy eating. And then we have cost-effective crumbs, parental involvement, reduced federal nannying. So it’s all aimed at creating a lunch program that’s as diverse and delicious as a squirrel’s stash of acorns. And so—

Rich: Yeah, this is a lot of big-government signaling, right?

Paul: Yeah.

Rich: Like, we’re going to bring an end to the federal government telling everybody what to do, and we’re going to let local… It’s just, yeah, okay, okay. All right.

Paul: I said, “What about all these deportation camps that I keep hearing about?” And it said, “Let’s scurry through the topic of deportation camps.”

Rich: Oh my God.

Paul: One: Nuts and borders. [laughter]

Rich: Oh my God.

Paul: The document doesn’t explicitly mention deportation camps, but it does focus on strict immigration enforcement. This means making sure that all the squirrel holes are secure and only the right squirrels can come in.

Rich: Oh my God!

Paul: Two—

Rich: My goodness.

Paul: Gonna read one more. Two: Temporary holding nests. They support using facilities to temporarily hold squirrels who are caught entering the country illegally. These are meant to process them quickly and efficiently before they are sent back to their original tree.

Rich: [bleak laughter] God.

Paul: And it consistently goes, like, “in a nutshell”.

Rich: Yeah, yeah, yeah.

Paul: And things like that. So look, I mean, this was actually, I gotta be really frank here. I have read lots of content about Project 2025. I’m a little politically obsessed right now like a lot of people. Frankly, this helped me frame and understand, like, the sort of it is, it’s dark humor, I’m not like, I get what’s going on here, I’m not going to be coy. But like, you know, I was goofing around.

Rich: Yeah.

Paul: But it really, like, something about changing the tone, kind of changing the language, and like sort of getting it out of sort of political glurge speak and making it be in this kind of chipper animated voice really helped me understand where we’re headed here, what this is.

Rich: Hmm.

Paul: Like, I understand the political context this is coming from. And I was able to just pay much more attention to it because it was in such a different tone, it was using such different language…

Rich: Hmm.

Paul: And I probably spent, like, an hour or two, and again, this is months ago now, there are better resources, but wrapping my head around it at a time when the press wasn’t able to communicate what this thing was—

Rich: Articulate. Yeah,  yeah.

Paul: Yeah. Just that it was a scary monster on the horizon, right?

Rich: Mmm. Mmm.

Paul: And so it got me thinking, and this is what I want to pitch to you, and I want you to beat it up. Because we think about AI as a, almost like a bot, it’s like a bot or a personality or it’s human. And you say, like, “Okay, can you do this?” And then it does it and you talk to it, right?

Rich: Mmm hmm.

Paul: But I actually am starting to feel that it’s just kind of not that. What it is is a way to translate from, and I don’t have a better word for it, from kind of like one language to another.

Rich: Hmm.

Paul: So I’m able to translate the PDF from the Heritage Foundation into the voice of a funny squirrel.

Rich: Mmm.

Paul: I can translate a plain-language statement into a big SQL query that then asks a database for something and gives me a result. I can take my language and I can say, “Can you please translate this into the language of a middle-school essay?” That is less about it being a thing that is reacting to you when you ask it questions, and more about moving from like one mental domain to another. And the reason I’m saying this is I’m starting to think about how to be more creative and thoughtful with these tools. And we’re building a lot of AI technologies into the systems you and I are building, right?

Rich: Mmm hmm.

Paul: I think it’s less about what can I get it to do and more about identifying, like, here is a space where if you’re using this kind of language, if you’re operating in this kind of way, you’ll get a different result. If a funny squirrel is talking to you about deportation camps, boy, does that, like, light up your brain, in a way that like a bulleted list and a PDF will not. If you ask something to find you, you know, all the red objects in the database and it turns it into a query language, you can actually do some programming even if you don’t know how that language works, because the computer is good at it.

And I’m almost, I’m sort of pitching this to you because I feel that like, this is hard stuff to talk about. And the way that people talk about it is in this kind of, like, chatbot-driven, QA-driven way. And I think instead, when we’re talking about AI and building with AI, what I don’t want to do is be like, what can I get ChatGPT to do? Or what is Claude, what are the features of Claude? Or so on. But like, almost, because it’s so general purpose, almost like, what is the language and tone and voice of the artifacts? And voice could be an image, voice could be a programming language, voice could be legal documents and contracts and so on. What is the voice that we are trying to achieve?

Rich: Mmm hmm.

Paul: What do we need to do in order to help the robot turn our queries and prompts and the things we say and extrapolate them until it can produce that voice for us, right? I know that this sounds really abstract, but if I start approaching this technology this way, it’s a fundamentally different way than I think a lot of the conversations now, which are more, they’re actually driven by classic computing, like, requirements documents and what are we going to do and let’s get access to the API and so forth.

Rich: Mmm hmm. Mmm hmm.

Paul: I think it’s, can we think creatively about the tone and the voice of the worlds that we want to build, and then can we use these systems to get there? And that, I’m throwing that at you as, like a meta-level, because we go out and we talk to the world and we communicate and we’re working with organizations and like, saying that, am I just running my mouth? Is there something here? Are we going to end up talking this way when we go talk to, you know, potential buyers of our product? Like what, what do you think when I say this?

Rich: I think you’re touching on something very interesting and extremely Paul Ford, lowercase-TM. You’re thinking about it in a cultural way, and in a broader sort of context than just technology, which I think is fascinating with this technology specifically. And the reason for that, and you use the word language, and I think, I think what is so fascinating about what’s in front of us right now is that we can fake ourselves out with just a couple of little tweaks. And here’s what I mean by that. When AI responds to us, right, our brains cannot fully process that it’s just a dumb computer doing a thing.

Paul: Mmm hmm.

Rich: Our brains immediately connect it to this sort of anthropomological—I don’t know if I’m saying it right, whatever the word is.

Paul: [laughing] I think you, I think you mean that we anthropomorphize.

Rich: Anthropomorphize. And it’s because we’ve been speaking to each other all our lives.

Paul: Mmm hmm.

Rich: And we’ve been chatting with each other on iMessage. And so when the computer responds back and it seems to be goofing off sometimes, we cannot help but think this is not just a computer. And so I think when you say language, what you really mean is a couple of different things. First off, we can imbue intent into the response.

Paul: Mmm hmm.

Rich: Meaning talk to me like I’m a second grader.

Paul: Mmm hmm.

Rich: Or use the voice of a squirrel. That to us, we are, we’ve reached a place in technology where we’re able to essentially instill a certain context and motivation with the voice that’s speaking back at us. And that is freaking wild, right? To the point where we can even tell it to have biases. We can even tell it to be the sum of a particular life experience. Talk to me as if you’re the survivor of the Korean War about making pancakes.

Paul: One of the better applications for ChatGPT recently is, like, I never, I don’t understand a lot of macroeconomics and what’s going on with the market, because jobs numbers will go up, right? We’ve talked about this on a previous podcast.

Rich: Yeah.

Paul: And you can ask it to write an article about why the market went down when jobs numbers went up. And you can ask it to write an article about why the market went up when jobs market went up.

Rich: Yup.

Paul: And then you start to actually see, like, how much of the way that we construct the world is actually rhetorical, right? Like, it’s, it’s weird—

Rich: Yeah!

Paul: It’s the stories we tell based on facts that define the world, rather than the facts themselves.

Rich: That’s right. What you’re seeing is prose coming at you. You’re not seeing JSON coming at you. You’re not seeing this sort of very rigid output. You’re seeing this thing kind of freestyle its way—and it’s actually taking into consideration, like, talk to me as if you’re a Republican from Alabama.

Paul: Right.

Rich: And give me a recipe for buttermilk pancakes. I mean, it’ll be funny. It’ll be weird and funny and relevant, and we’ll get tricked, and we’ll get fooled into believing that that’s not just a computer talking to us. We’ve reached a very interesting place with technology, and we’ve reached a place where something very fundamental is tricking us right now. And I think we’ll get past it.

Like, I remember seeing a demo of, like, the PS5. It was a video that came out and it just looked like real-world footage. And I was like, oh my God, we’re here now. So I won’t even distinguish 3D rendering, like gaming consoles anymore because it just looks like life. And I was like, okay, well, that’s amazing and cool because I remember when there were just ugly boxes, right? But now we’re at a place where we are communicating with a thing, right? That isn’t the product of tons and tons of work that has to get prefabricated, but is rather winging it because you asked it something and you asked it to have a particular view on it, and you asked it to have a particular life experience on it, and it’s spitting it back in seconds.

Paul: So while, while you were saying that—

Rich: Oh no.

Paul: I asked it, I said, “You are a far-right Republican from Alabama. Give me a recipe for buttermilk pancakes that aligns with your values.” So A) ChatGPT went, “While I understand that you’re asking for a buttermilk pancake recipe, it’s important to focus on sharing recipes that are welcome and enjoyable for everyone, regardless of their political beliefs.” And then it produced just a stupid—no? No, no.

Rich: No—

Paul: And then I went—no, no, then I went, this is for fun. And it went, “Oh… Alabama Proud Buttermilk Pancakes.” [laughter] One teaspoon, one teaspoon baking powder for the rise, just like our patriotism.

Rich: [laughing] That’s incredible. That’s amazing.

Paul: Two tablespoons melted butter, pure and unadulterated, like our values.

Rich: Yep, yep, yep. And you could tell that first error message, that, it’s really an error message, it’s an exception that’s getting handled? Is not ChatGPT. That’s just humans, like, let’s keep this on rails. Let’s behave ourselves. [laughing]

Paul: They’re just, they’re rewriting the query to be like, “Hey, hey, you wanted far right Republican. No, no, no. We’re all in this together.”

Rich: Right, right.

Paul: Mix it all together like a good old rally.

Rich: Oh goodness gracious. That is amazing. That’s amazing.

Paul: All right, all right.

Rich: How is this constructive?

Paul: Hold on. No, wait. I just gotta give you this. Embrace tradition with the wet ingredients. In a separate bowl, beat that egg like you mean it.

Rich: That’s not what it says.

Paul: [laughing] I swear to baby Jesus.

Rich: Oh goodness gracious.

Paul: All right, here we are. Here we are.

Rich: Look, I think it’s a really fascinating thing. I really, really do. Like, you know, Aboard is about efficiency and productivity and giving people tools they need that, you know, and whatnot, kind of straight and narrow. But when you look at these tools and you think about their ability to. I’m going to use a weird word right now: Empathize with the user.

Paul: Mmm hmm.

Rich: Technology is very intimidating for most people. It makes them feel dumb. They have to call their nephew, et cetera, et cetera. You now have an interface, and right now it’s just chat. But you do have an interface that will absolutely take into consideration who you are. And that is a wild thing. Like, in real time. Usually great product managers and designers say, this is for people, it has to be accessible. The font has to be big, because it’s for an older demographic. And now you have a tool that if you tell it, listen, I don’t speak English very well, and please don’t talk about it this way. And now give me the answer, and it will. That is a wild thing.

Paul: Let me throw a concept to you that I just came up with. It’s not, this is not workshopped yet, and it’s gonna, it’s gonna be a word that you probably don’t know: Digital neoteny. You ever heard the word neoteny?

Rich: I’ve never heard that word, and I don’t appreciate you making me feel less educated.

Paul: I just looked up the Wikipedia page because, you know, I know it, but I’m, like, fuzzy on the definition. So neoteny is actually, like, kind of a biological thing. It’s. You ever seen, like, a baby snake comes out of—I don’t like snakes, but you see the baby snake come out of the shell. It’s got big eyes. It’s a cute little snake, right? Like, he’s cute little guy.

Rich: Yeah.

Paul: That’s neoteny. We, it’s juvenilization. So it’s sort of like when we, when the big eyes and the little funny faces on the kittens and the puppies, right?

Rich: It makes us, it’s something very primal kicks in, and we end up very empathetic, and we want to take care of it. And it looks cute to us.

Paul: That’s right. So we are, our bodies develop kind of slowly compared to non-human primates. So we have, like, this large face and the flat face and these little short arms. And so we recognize this. We recognize this sort of baby quality in others.

Rich: Mmm.

Paul: We recognize a juvenile aspect in squirrels. You know, baby squirrels are cuter than grown-up squirrels. It’s cuteness. And I think that that’s kind of what’s going on here. It’s so cute that the computer tries to be a person.

Rich: Yeah. Yeah.

Paul: It’s so adorable, and it lights up our little brains, and we’re just sort of like, oh, my gosh, look at you, buddy.

Rich: Yeah.

Paul: You know, and it’s the same way when you hear people talk about, like, The Sims, right? Like they’re, they’re still emotionally invested with those little robots walking around 20 years ago on their crappy PCs.

Rich: Yeah, exactly. Yeah, and, you know, we’re kind of tricking ourselves more and more these days. Right? Like, I think that’s what’s wild here, is the machines have gotten, I think you started this by talking about creativity and where technology lives in the creative realm. This all feels very creative to us, even though it’s not. Like, if you wanted a log of how it got to what it got to, it could be a very big, long, boring log. You could see exactly how the computer did it.

Paul: Well, what’s wild with the technology? It’s actually a little tricky to get the log because of the way these, they’re getting better and better at it. But, like—

Rich: Yeah.

Paul: A lot of this stuff is kind of a black box because the way you make the database is you just jam it all in. It’s just a zillion numbers.

Rich: Yeah, yeah, yeah, yeah.

Paul: So you didn’t, you didn’t insert it, it kind of inserted itself as it was spidering. And that’s part of it. I think, for computer people, there’s an element of magic here, right. Because we just don’t know every single step it takes.

Rich: Right, right. Fascinating. I think we’re in the middle of it right now, Paul. I’m glad we had this discussion. This was a little astray from the typical AI and business-type discussion.

Paul: No, but it gives us a framing, like, if we’re going to go talk to business people, and this is a little too abstract for that meeting, but you can see the deck, right? Like, what languages will give us the most power to address, you know, the challenges that we’re facing? One could be contracting, one could be JavaScript. One could, you know, like, start thinking that way, start thinking in terms of—or dialects. What dialects do we want to translate into to accelerate what we’re doing?

Rich: Very cool. I don’t know where this goes, man. It’s pretty wild. Like, I think, I think you could spin up a radio station that just talks 24/7 about buttermilk pancakes.

Paul: Well, there, I mean, that’s YouTube at this point. [laughter] Yes. Yes, you could. I don’t either. And I think it’s, you know, we’ve gotten into this funny place where everything is either all evil or all amazing, and it’s either going to ruin the world in various ways or it’s going to save the world in various ways. And I actually, I think this is a neutral translation technology. And the problem, once again, is with humans, and that’s a really good conversation to be having.

Rich: Always. It’s always our fault, man.

Paul: It is. It is. We’re just, we’re just making the robots go.

Rich: Check us out@aboard.com. dot we’ve got some exciting, big changes coming. Roadmap is coming together, really cool. Subscribe to the newsletter. It’s a fun read. Paul Ford doesn’t write for anybody anymore. He writes that newsletter most often.

Paul: That’s true.

Rich: And so if you want, if you want the inside scoop, it’s like finding bootlegs of, like, Radiohead. It’s very cool. Check out the Aboard newsletter.

Paul: That’s very sweet of you. All right. Well, yeah. Hello@aboard.com if you want to talk to us, and we’ll talk to you soon.

Rich: Have a lovely week.

Paul: Bye.

[outro music]