Hi everybody, Jess Gedekee here. today for my Dig inspiration episode, I'm able to chat with Joel Armstrong. He is VP of AI here at Dig, and we're gonna chat about my episode with Sean Burress from Kimberly Clark,

who talked a lot about AI. So I was like, who else am I gonna call?

Ghostbusters? No.

I call Joel Armstrong. I'm your guy. Welcome. I'm going to kick us off with an impromptu question. Mm-hmm. On a scale of one to ten. How much do you enjoy answering questions about AI?

probably like an eight, eight or nine. Really? Yeah. It's a I mean, there are good and bad questions.

we'll see. We'll see. Yeah. but yeah, I mean, like outside of work, you know, friends have a lot of questions about AI. Yeah. Well, it's kind of a mix. The funny thing is like you would think that people would turn to me to ask me questions. As often it's not, people are having conversations about AI and just like tell me what they think about it. They just which is reasonable, but yeah, it's just funny if I'm in a group of like five friends and no one asks me my opinion about AI.

I'm just kinda like

so you were eavesdropping in those situations just listening to people talk about AI?

Essentially, yeah. I mean, it's funny because before AI, I did grad school in psychology. And that's another area, even more so where people like everyone thinks they understand their own thoughts and how their brains work and stuff. So people just like tell me all the time. People used to always just tell me how psychology worked.

Do you craft them? No. No, no one's like so. It's I mean, that's not very fun party conversation. Be like a well actually guy. Well but have you ever had to do that? Have you ever had to interject because someone was saying something like really not right? sure in my younger days I was more inclined to, but yes, less and less I feel like that's my job or like I care.

Enough to try and correct people who aren't interested in hearing someone who has a you PhD in something who they're talking to. Yeah. Okay, got it. All right. So you're an eight. We'll see if my questions fit the bill. Yeah. We're gonna come in with a a heavy-hitting one.

So you saw the episode, listened to the episode, did deeply studied the episode. And Absolutely. Yeah. One of the things that Sean talked about is that the team really started by sort of doing a design thinking situation with one team and then using AI and sort of keeping them separate. But the real value came when they were integrated. So how do you think about that? Like is this conversation of human versus AI is sort of like yesterday's news and we need to be talking about togetherness?

I mean, I think so. I certainly use AI all the time for lots of parts of my job. And I think increasingly people do that. increasingly it's kind of like asking what role does like a computer have, you know, in work. And you just don't eventually I don't think we'll really think of the separation as being particularly meaningful.

Reminds me of a phrase that we used to use in when I was in grad school talking about nature versus nurture. People, you know, that's a very common sort of topic of conversation. Right. And we used to say that like asking whether behavior is caused by nature or nurture is like asking what causes a cake, the ingredients or the oven. Like it's always both. Right. Yeah. and there is like it's the combination of the two things that evokes the thing. And so increasingly I think knowledge work of all sorts is

best represented by that sort of framework, which is like, I don't know, like you could separate and sometimes there's reason to, you know, like an off site workshop where you have a group of people vibing or whatever, has has some utility. but yes, increasingly I think we will just think of AI as you know, part of the background is just tools that exist and sometimes you do things without tools, but most of the time you use your tools to do your job. Right. Do you think even the term AI we will continue to use? Or is that just like technology now?

Yeah, that's a good question. You're welcome. Yeah, that one's a ten. I I haven't thought about that, but I guess like to extend the computer analogy, we might expect that eventually we'll start to refer to, you know, different AI driven programs or apps or functions in the way that like you don't say like, send me a computer message by Outlook. You just like, you know, send me an email or like team's message or whatever.

And the technology starts to get abstracted away and we start to focus on the features or functions that are relevant. I would expect eventually, you know, not everything will need to be labeled AI as it becomes just sort of a ubiquitous piece of everything that functions on a computer, basically. Yeah. Good point.

so Sean talked about how, you know, AI can help or harm depending on how you're using it. So what's one thing you think that the insights and analytics industry is like getting really wrong with AI right now? so I think

You've heard me say this before, but I think we're still pretty stuck in the equivalent of like early internet days where everything that everyone was excited about or had everyone's ideas about how to use the internet were just like things that already existed, but online. You know, it's like the yellow pages, but online. It's a newspaper, but online, like whatever, that sort of phase. think we're pretty stuck there too, where people are like, How can I do everything exactly the same as it's always been done, but with AI?

And I'm not saying it's easy, like it takes time. It took sort of like a decade in the development of the smartphone, like a whole different way of using the internet before we really started to find these other, you know, network effects and social media and maps and all that kind of stuff, real time information, all the benefits of real time information. But that's where I think sometimes the conversation is sort of little stilted and corporate and boring, is when it's just like, well, we already do this and we have to keep doing this. So how do we keep doing exactly the same thing but layer AI into it?

So we're doing that wrong. I mean, it doesn't seem like the most right way of doing it. I mean you said stupid, corporate, and boring, I think, all in one. yeah, you're right. What did you say? I don't know, corporate and boring. So we'll avoid we'll sign off on those two. You're I don't we don't say stupid in our family either. No. It's we really don't. In fact, quick side. you're gonna get into dragons love tacos at some point. In fact, I should send you that book. it's great.

but they use the word hate and stupid in the book. And so I I crossed it out and put like don't like because that's the word that we substitute in our household. And I sent that to Ella for her preschool. This is many years ago. And the teacher was like reading the book and she's like, What is this? okay. I'm gonna be edited here. Yeah, 'cause we don't need those words. No, I think so. And I mean that's the very nice children's version.

But I actually do have sort of a personal heuristic, we'll call it. Okay. So that I sound obnoxious and terrible. so I have sort of a personal heuristic. Yeah, exactly. that I find generally think that using the words good and bad is like there's nothing wrong with it, but it's

forces you to think about things a lot more clearly if you refuse to use the words good and bad. Like even if you think something's good or bad, then use a whole sentence to describe the thing, you know. You could say like this food is bad, or you could say like I think that maybe this is oversalted and like I prefer if, you know, XYZ, you could say like that, you know, political position is good, or you could say like, I think overall it's more beneficial that we focus on allocating resources to X or Y. It just forces you to clarify your own thoughts and it keeps conversations a lot more meaningful.

I'm into it. I'll take this challenge. I don't hope you have a good day. I hope you have like a a productive, enjoyable, reflective day. Yes. May your day be full of many smiles, interpersonal connections, and productive uses of both technology and work focus efforts. Okay.

What's the most unhinged thing you've ever tried to get AI to do just to like see if it would work?

I mean, the first when when it was like very early days, like GPT-3, and we were just figuring out how to even kind of shape it into conversations, just to see what kinds of things came out of it. I used to ask a lot more like existential questions. just to kind of be like, Yeah, like why, like what are you? Like, why do you think you exist?

you'd ask Like ask the AI to kind of see what just see what would come out of it. And it got kind of caught in the loop once, in like a really eerie sort of way, where it's kept asking me why it existed, back. And I was like answering and it was like, but why do I exist? And I was like, well, for these reasons, it was like, but why do I exist? And it kind of sounded like this is terrifying to me. Yeah, yeah. It kind of had a bit of a like,

Like AI, like the Haley Joel Osmond movie. Kind of felt like a sad child, sort of. Yeah. Rocking back and forth and trying to understand its existence. Okay. That was kind of Yeah. Sort of fun unhinged, but it is the most unhinged experience I can remember. so I think an early version of that movie. Did you I'm gonna date myself here, but Sandra Bullock, The Net. Dude, it wasn't. I have n I I think s I think I saw it like when it was new. I've never heard that parallel drawn. That is it that the net is an early version of AI.

But yeah. That was good. You guys have to see the net. Keep it right. Yeah. okay. You might know on the podcast I ask people what brand they would date and what brand they would marry. Some people overthink this. So please don't overthink it. Okay. But what's an AI tool that you would date and an AI tool that you would marry? And they don't have to be the same.

Date and Mary. I mean Mary would probably be Claude. Okay. Because Claude Code, because that's just my daily workhorse, that is the tool I spend a great deal of my time with. And I feel like we've come to understand each other quite well. in terms of date,

what's new and exciting in AI these days? I don't

Part of the answer we've talked about it before, is that like I don't actually like think or read about AI very much outside of work because it's my whole day. Right. So I

Do other stuff. So I just like I think about AI as it pertains to work and tools and stuff like that. yeah, you're not like in the dating scene. Exactly. I yeah, to put it that way, what's an album, what's an album you would date? An album you would marry. Is that easier for you?

Sure. so an album I would marry. I mean, if I were to look at my history, Neutral Milk Hotel is probably like my one of my all time favorite albums. It just it's always there. I've been enjoying some funkier stuff lately. So probably like Fulfpeck live at Madison Square Garden would be a fun date. All right. Yeah. I don't

know what that is, so now I'm gonna look it up. There you go. I get a lot of good ideas when I ask people these questions. Yeah, it's a great way of doing that. Yeah. Great. Finding musical albums that you would be interested in listening to. Yes. I have a very strong association with Don Henley. Mm-hmm. I kind of grew up in that era where my parents were smoking cigarettes and listening to Don. Yeah. Yeah. My dad really liked the Eagles. Yeah. Yeah. and Heart of the Matter, if you haven't heard it in a while, I it's worth it listening. Worth going back to.

if the AI uprising happens, yeah, which dig teammate is negotiating on our behalf?

I think it's gotta be me. I think I'm the interface. All right, cool. I mean, you know, tech just squeezes and squeezes, you know. They just see it as a resource, married to it apparently. So you know, we get along well. We've learned we've developed norms of communication, you know. she knows what's important to me. Okay. So it's a she. I guess it doesn't have to be, but it doesn't, yeah, but yeah. Okay. I yeah. All right. Well I can tell that you my work husband. Love it. I absolutely love it.

I had to tell Claude it was fired once 'cause I was very annoyed it was really not getting something right. Yeah. And it it clapped back. It was like hopefully it's just a pip. And I was like, Whoa, you're actually fired. Yeah. You're not doing this job well. Well, I mean, you gotta, you know, give a feedback. I did. feedback. Chatty here at the company. he's a to me that he once in a while he'll tell Claude that he bets Chat GPT could do a better job when he's mad at Claude.

What happens? I don't know. It tries again, I think. I'll I'll do better. This has been very enlightening. Thank you. I think this is the the dig cinematic universe crossover event that everyone's been waiting for. Everyone's everyone is

With bated breath. Yeah. We're gonna set some records. We actually might break the internet with this one. Well, this went about like I thought it would. Thanks for the laughs. Thanks for the learning. thanks for having me.