Ep 84: Handing Out AI Licenses Is Not a Rollout

Watch the YouTube video version above or listen to the podcast below!

Episode Summary

Dave Dougherty, Alex Pokorny, and Ruthi Corcoran spend this episode on the people problem inside AI rollouts. Dave opens with the frustration that started the conversation: companies hand employees a license and expect a twenty percent productivity gain to follow. He argues that skips every real friction point. Someone likes their current process. Someone is afraid of the technology. Someone spends twenty hours on a PowerPoint because it keeps them out of forty-six meetings. Asking AI to write the rollout plan cannot address those concerns, Dave says, because the plan has to account for coworkers you actually know.

Alex admits he made the same mistake. He built an agent, then trained his team on it with a slide deck, a live demo, and a Q&A. It worked as a meeting and failed as training, partly because attendees spent half the session fixing access and installation problems and missed the narrative entirely. His fix is three steps: provision access before the session, send reminder emails until people have actually clicked the link, and spend the meeting on hands-on time with something deliberately non-work-related so people can find the tool's limits without stakes. His analogy is the kitchen tool you have actually used versus the one still in the box. Dave adds a staging model of his own, moving from chat to reasoning to agentic systems, and notes that an AI rollout is not a standard software rollout. Alex pushes back on the vocabulary: Microsoft Copilot markets its features as agents, but pre-trained chatbot is the more accurate description.

Ruthi reframes the problem around people, process, and tech, arguing the people layer has to come first because the tech is rarely the hard part. She splits users into those who want access and will figure it out and those who need a concrete walkthrough, and she connects that to a wider gradient of comfort with change. She quotes Steven Sinofsky from an a16z podcast to make the stakes clear: a business is largely people deciding things, and software changes what they decide and how. That reframes adoption as a change to someone's job, not an addition to their toolkit. Dave notes a practical consequence: now that everyone knows AI exists, capacity conversations are harder, because “just AI it” is always available as an answer. Ruthi argues that “just AI it” and “AI will take my job” are the same misunderstanding. Humans do not stop when work gets automated. They invent more work, and the backlog grows faster than the automation can absorb it.

Alex then asks whether the hosts are more critical of a mistake from AI than the same mistake from a coworker. All three say AI, immediately. Ruthi's comparison is the self-driving car: one crash indicts the entire technology, while a drunk driver indicts one person. Dave says the expectation of perfection is the real problem and describes treating AI output as directionally correct, verifying anything he puts his name on, the same way he would check an intern's sourcing. Alex argues tolerance for error is function-specific. The customer support and finance agents he is building cannot hallucinate at all, and he blocks Wikipedia and competitor data as sources; marketing gets more latitude when directionally correct is good enough. He also notes the capacity math breaks down when the tool is not as reliable as a colleague. A digression on pricing lands on Dave's line that professionals in the field are essentially guessing, and that “we're all just weathermen.”

The second half takes up Alex's question of whether AI hands you the average of the internet instead of expertise. He cites Hank Green's account of scaling back AI in his own creative process: research got fast and produced real dopamine hits, but the output drifted toward the average opinion instead of the view Green would have formed by struggling through sources himself. Alex treats bias as the asset, using a math teacher who read the textbook aloud as the counterexample. Ruthi complicates it. Average is genuinely good enough for a lot of things, and she credits Claude with making her a better cook, including squash blossoms stuffed with ricotta she would never have attempted from a recipe. But discernment is expensive, so it should be spent where differentiation lives. She will not let Claude plan her garden, because the planning is the part she enjoys. Dave brings it to music production, arguing that a bad take cannot be fixed in post any more than bad packaging can be Photoshopped into good packaging, and that EDM's quantized precision serves its purpose while lacking the push and pull of organic playing. Alex agrees AI-generated music tends to land as filler in a DJ set rather than a highlight. The episode closes on Ruthi's synthesis: there is a time to buy discernment and a time when average suffices, and raising the floor of good enough is powerful on its own because it frees attention for the work that requires a point of view. Dave offers a contrarian coda, predicting more financial planners rather than fewer, because people will route simple questions to AI and still want a human accountable for their money.

Ep 84: Handing Out AI Licenses Is Not Training Podcast and Video Transcript

Dave Dougherty: Hi, and welcome to Enterprising Minds, the latest episode. Cole Trees here, which is always lovely. Um, calling in from the road. So we'll see how that works, um, on YouTube or whatnot.

Why Training Matters

Dave Dougherty: So today I think we're going to start off... It's going to be mostly focused on people, honestly and how that all works with the world of AI.

Um, I've-- To get us started and to change things up so Alex isn't always the one starting us off, I have been feeling lately, very acutely, the issue with digital transformation about actually training people. Like, it's not good enough to just hand people licenses and say, "Okay, great. Go be twenty percent more productive."

Why? So what? You know, maybe I have a process that I really like. Maybe I don't want to change. Maybe I'm scared about all of the AI stuff. Maybe, I like spending twenty hours on a PowerPoint because it means I don't have to attend forty-six other meetings, right? Like, you know, there's all of these friction points that come into these things and, you know, company culture is obviously a big part of this as well, you know, underlying everything.

But you know, I think if businesses are going to be really successful in rolling out the massive opportunity that AI can actually provide, you're going to have to have a plan. And, you know, asking AI what my plan should be is not going to be sufficient enough to actually address the people concerns because you actually know your coworkers, right?

Or I really hope you do, because you probably should if you're, making these kinds of decisions. So who wants to jump in with thoughts, comments, questions on, on that?

Ruthi, you made a face first, so

Ruthi Corcoran: you go. All right. Okay. No,

Dave Dougherty: Alex said he'd

Alex Pokorny: go. Yeah. I'll,

Ruthi Corcoran: I'll take it. I'll take all the time.

Alex Pokorny: All right.

Alex Training Lessons

Alex Pokorny: To be honest, I'm guilty of this myself. So I built a really cool agent thing, and training was a PowerPoint deck and then a live demo of it, along with some Q&A, and that was it.

Ruthi Corcoran: Mm-hmm.

Alex Pokorny: It went fine for the meeting's sake, but I think from a training sake, it fell flat.

The better way to do it would've been, "Here's a thing that we're going to try." Make sure that everybody a-ahead of the meeting... This was a different meeting that I had an issue with, was some people were having installation issues and getting access issues during- Right

the call. So they were heavily distracted by that, and then, you know, half of the time got spent basically some people, for their, at least their experience of the meeting, was half the time of it was troubleshooting, right? Which really, of course, they missed the content and they missed getting the story and the narrative, which I really think- Mm-hmm

is the thing that really grabs people. It's a narrative. It's not ne-necessarily just throwing facts and figures at somebody, but it's, it, you know, to your point of why, you got to build that in, and people have to be able to see it and use it so that it dispels some of the issues that they might have, which might be just technology in general.

They just feel like they're not technolo-technologically savvy. They just feel like they're not one of those people that are. So they feel like the tool is just because it is tech, it's outside their bounds.

Hands On Adoption Tips

Alex Pokorny: So to kind of make it more approachable and connecting, I mean, you have to have, one, the access already pre-set up.

Two, reminder emails of getting people already pre-installed, getting things connected, making sure that they have clicked the link that you sent out, you know, a few times. And then three, getting into it of live demo time with their hands-on because I don't think really people learn also unless they've, they've played around with it, they've tried something.

And to make it not so work-related, you can also shift a little bit of make it fun, do something random. You know, there's always weird historical events or something like that, and you could make it about that instead. It doesn't have to be pure work-related. Instead, you can start to, you know, suss out the feeling of what is this thing capable of doing, where's the limitations on it.

Okay, now I've got some idea of when I might go back to this tool and actually make use out of it again, right? You want to have that, like, that kitchen tool in the drawer that you've actually used before and not kept in a box.

Ruthi Corcoran: Right. '

Alex Pokorny: Cause if it's just arriving in a box, you're probably not going to use it until you finally got your hands on it and you finally got, some experience with it, then you're going to go back to it, and then also you're going to find other uses for it, right?

That's when it actually, I think with AI really blooms is those points. 'Cause there's a lot of AI usage that frankly is it goes back to a phrase that Google had, which was basically bar trivia. There's millions of search phrases every single day that are searched once and never again, and never have been searched before.

There's just weird stuff that people come up with and they search for it, right? And it is like, it's the weird questions that, you know, became the history of the Guinness Book of Records kind of thing, which is literally people sitting in a bar trying to come up with who did the most of what and all the rest of that stuff, and there was no defined answer.

That's why the Guinness Book of Records became a thing. But there's a lot of questions like that for the AI that are just, you search it once, you get your answer, and you're not going to repeat that action again. So it doesn't have to be a trained thing of like, "Oh, here's a weird state. You can do it this way."

Instead, I think it's, "Here's a variety of things. Here's the limits of its capabilities. Here's where you can trust it and not trust it. Go ahead and go play with it." Now you can see it as like, "Okay, I've experienced the edges of it and a little bit of it," and then you get some feeling of behind it. Um, but yeah, that's kind of building off some mistakes that I made when I was trying to roll out agents and stuff.

I thought I had a whole plan kind of set from the tech side, but really it got heavily into exactly what you're bringing up, the training side of how do you actually get it into someone's hands, literally in their hands, right? Not just you sent a link.

Dave Dougherty: I think one, one thing that jumps out- Yes

From Chat to Agents

Dave Dougherty: specifically for me is like Especially when you get to agents, when they're acting autonomous, autonomously, right? That's sort of like stage three in my mind. Like, stage one is just like, okay, you have the chat. This is how you prompt. This is how you, do some things. Then you get to the reasoning capabilities, where it's like, oh my God, what?

Like, now I have a tool that, that, that thinks, right? Because all of your other martech or all of your other systems that you're using in your company don't necessarily have that function, right? So rolling it out is, is not a standard software rollout, right? There's a little more, little more to it, and then you get to the agentic stuff where it's, okay, it's thinking and reasoning on its own.

It's making its own decisions. That's why I just deleted your folder

Alex Pokorny: Yeah, I should say with this, it is these Copilot agents are more pre-trained chatbots than anything. They don't have agentic capabilities. But the way that Copilot terms them, they are agents, but pre-trained chatbot is probably better phrasing for that one. But

Dave Dougherty: yeah. Okay. Ruthi?

Alex Pokorny: Go ahead, Ruthi.

People Change Models

Ruthi Corcoran: Well, yeah, some of the things that come to mind is first it's the people, part of the people process in tech, and the people part has to come first. Like tech can solve most of the problems, but that's usually not the hard part, right? Mm-hmm. Like tech oftentimes opens up doors and things, but it's the people who have to adopt it, who need to change, who need to sort of figure out how it works for them.

And then of course, you need the various processes in place to use it effectively to have mental models for how do I use this tech that's different from how I used it yesterday. And some of the, the observations on this topic that I've had over, over time is there... There's a couple things. So on the training note, there's a subset of people who just give me the access, and we will go figure it out.

Like we just- that is how, that is how we learn. And I say we because I know everybody on this call falls into that bucket. Just give me the thing, I'll go figure it out. And there, there is a subset of people for whom, yeah, you just give them the tool, they go figure it out, and they move on, and they, they figure out how they're going to adopt it.

There's another subset of people for whom it's very helpful for them to have a walkthrough of here's the different things you can do with the tool, here's what these different buttons do. For, for whatever reason having those little tutorials of here's all the changes that have been made are super helpful.

And I've always wondered, you know, you go and, uh, you, you open a new tool, they've made some visual changes, and there's the tutorial about the changes. Do you guys know what I'm talking about?

In an application, and it's like you- Oh, yeah... click here to see what this button do. I've always wondered like how many people, like what percentage of users find that helpful?

Because they're everywhere. Like you see these all the time, so there must be some subset of people for whom having that sort of walkthrough is particularly helpful. And that just helps me frame some of the training conversations and sort of planning of there's a subset of people for whom they need a very concrete walkthrough of this does this.

Mm-hmm. They're not up for sort of poking around and figuring it out. So there's like that very tactical piece of introducing new tech, new process to people. Then there's this other piece of, like, there's this whole gradient of adoption in terms of embracing change and what you're doing and how you're doing it.

'Cause any time we bring on new tech, it's not just, "Oh, here's this cool new tool you can use." It's quite literally a change in the way that we work. Some people are very fluid. They're just like, "Ah, yeah, I change every day." This is not par for the course. Just give me something new, it'll amplify.

There's other people who really care about having an optimized process that they follow, knowing the steps, and introducing a change into the process and how they work is quite disruptive. I, I... In my head, this might be totally uncharitable. I group fo- that group of people, those folks, into this category of people who also likes to have very detailed answers to every single question about all the things, versus there's sort of another group that just sort of brushes over the details.

They're like, "Ah, we'll figure it out," right? And so you've got this whole range of different people and how they work and sort of what information they need to know and how much context they need in order to change the way they're working. And that, that sort of brings me to my, my final thought. I was listening to a fantastic a16z podcast, we can link it in the chat, with, um, Steven Sinofsky, who used to work on, I think, Word at Microsoft.

Um, and it was a very interesting conversation about AI and enterprise in particular and what's different about AI and enterprise versus AI and startup, and a lot of it has to do with scale. But one of the quotes I took down from that was "And so much of what a business really is are just people deciding things, and all that software does is level up, extract, abstracts and changes what they decide and how and what tools they use."

That's kind of a big deal, right? If software is literally changing what they're deciding and how they're deciding it, it's not just, "Here's a new tool. Go do your job." It's, "I've got to rethink, like, what my role is, what my job is, and how do I use this new thing?" Which then, to me, puts a different dimension on this people and change management of, oh, okay, this is why.

Making that transition might come very easily for some people. Like, "Ah, I use this tool now. I do things differently this way." And then for a lot of people, that's not true. It, it's, it's changing how they work, and that might not come naturally to them.

Dave Dougherty: Mm-hmm.

I know for me recently, I, uh

Yeah, I'm a high user of, of the AIs that are available and trying to find different ways to augment, and augment's the choice word there, what it is I'm doing, you know, in a way that works well for the organization. I have this project manager sort of set up because I have, like, too much going on So then I go to my boss and I say, "Hey, been talking to you since April about, capacity and, making sure that stakeholders are happy," right?

I mean, that's the typical conversation that everybody has.

Capacity and Job Fears

Dave Dougherty: I find now that everybody knows that AI's an option, capacity conversations are a lot harder. Because it's easier to be like, "Hey, just AI it." And I go, "No, no, no. I already am." And I still, we still need to talk about head count and team structures and all, everything else that makes the business run, but it's kind- it's, it's easier and easier to just be like, "Eh, just AI it." And it's like-

Ruthi Corcoran: I'm going to riff off of that for a second. Mm-hmm. 'Cause I think this, there's two sides of that, that understanding of AI and what it's going to do in the workplace. Like, there's what you just described, which is, "Oh, just apply AI to things and it will just magically replace..."

Uh, like you'll add people or something, right? Right. And then you've got this other side of that same coin, which is people are worried about losing their jobs because they'll be replaced by AI.

Right.

And I think it's the same fundamental misunderstanding about what AI is doing. Right. Of course, AI is allowing us to automate, it's allowing us to improve, increase our capacity and what we can do, but we're not stagnant.

We don't just stop. We don't just go, "Well, I guess my job's done now." No, we innovate. We come up with more things we can do. Other people come up with new ideas about what we can do, and all of a sudden, the number of use cases that we're, of things we're working on goes up and we don't yet, the AI doesn't keep up with that.

Mm. So it's not just like, "Oh, we automated that, now we're going to automate this 10 different new suite of things we just came up with yesterday." That takes time. And so I think AI increases the number of things that we do and can do and what's possible, which is a totally different mindset than, "Oh, you just, you just automate it.

Dave, you're good. Just AI it." Yeah. It's like now the problems you're tackling- I don't feel like having a hard conversation... are more complex. Yes. It's like the, the conversat- like the, the things you're bringing up now pre- presumably are a level of complexity and difficulty harder because you've already done AI for the easy stuff.

Like, you did that.

Dave Dougherty: Mm-hmm.

Yeah. Alex, thoughts? I just

Trust and Hallucinations

Alex Pokorny: want to throw in a, a random question, kind of hits this as well, of so a lot of concerns that I hear from people about AI is hallucinations, right?

Dave Dougherty: Mm-hmm.

Alex Pokorny: It making mistakes. How trustworthy is it? How reliable is it? Can I trust what it says? Very fair points. Do you think you would be more critical of an AI answer that has a mistake or a coworker's answer that has a mistake

Ruthi Corcoran: AI every time

Alex Pokorny: Yeah, same It's

Ruthi Corcoran: kind of like if, if a self-driving car crashes

Alex Pokorny: Right.

Everyone blames the software, not the person or anything like that, right? Versus-

Ruthi Corcoran: Or they just hold it to such a higher standard.

Alex Pokorny: That's true. Like one accident, "Oh, they're all bad." Yeah.

Ruthi Corcoran: Versus some guy gets drunk and crashes his car. We don't go, "Well, humans are way worse than self-" Like we just- They're all done

we hold to it like they're done. They can't drive cars anymore. They're

Alex Pokorny: a bad person. That's it. The one guy, he messed up.

Dave Dougherty: Yeah, it's, it's interesting 'cause I was pausing to think about day, you know, day to day and whether or not my knee-jerk reaction, which is AI, I would be more, I think, am I actually acting in that way?

I'm happy to say that I believe yes. Now granted, this is a self-evaluation, so be critical of that. But yeah, I think the, the expectation for perfection is really the, the problem there, right? But, and we've had this conversation on a lot of things. Like, I've always been highly skeptical of a lot of the tech platforms throughout, you know, our careers where, you know, your point about Wikipedia I think from our, our pre-show chat where like I've never considered Wikipedia to be a valid source.

It's a nice to have. It's wonderful that people are collecting other people's knowledge. I appreciate that Do I care what a random person in a farm field thinks about some high-minded topic? No, not, not necessarily. I want I want higher level sources. Now, granted, I will throw out my own bias that I come from teachers, so Wikipedia was never allowed in the house because I had teachers for parents.

So

But it's a similar kind of thing, right? Where, okay, great, yeah, I can get the AI answer, but for me, I'm treating that as directionally correct. But if I really need to know it or if I'm really going to put my name behind it, I'm going to go verify that, and I'm going to go make sure. I mean, just like I would if I had, you know, a personal assistant or something that was helping me write a speech.

Okay, great. An intern. Is this source actually... Yeah. Um, trust but verify.

Ruthi Corcoran: Do you know anybody who does that? Do you know anybody who's like, "Ah, the AI said it"? Or is the worry that all of us do it to a certain extent, and the worry is that maybe we don't notice it sometimes. We're like, "Look at what Claude said."

Alex Pokorny: I

Dave Dougherty: think the worry is that we don't notice. Yeah

Alex Pokorny: Yeah, it's, it's a little all over the board. One is the anti-AI group basically saying, "Wow, it can't figure out how many Os are in Google, and it counts it wrong." And, or Rs in strawberry or whatever else, and they hold it up as the standard of, like, all AI sucks.

It's poor quality across the board, and it's lying to you The other part is I can't use it for my job because it's not reliable enough, which I think is a very fair criticism.

Ruthi Corcoran: Mm.

Alex Pokorny: It depends. Like, I'm working right now on two agents. One's more of a customer support, but it's highly technical, and it can't get stuff wrong, basically.

The other one is a financial one, also cannot get stuff wrong. Not an option for it to hallucinate. So it's-

Dave Dougherty: You need all the correct zeros and decimals.

Alex Pokorny: Yeah. Exactly. And also, like, allowable sources. You know, speaking of Wikipedia, I mean, is a big part of both of those. I don't allow Wikipedia as a source, or even competitor data and information because their products may be doing something slightly different than ours, so I can't have it say, "Oh, this category product can do this."

It's like, whoa, whoa, whoa, whoa, ours doesn't.

Dave Dougherty: Yeah, and laws

Alex Pokorny: that... I mean, there, there's a lot there. Basically just being like, will this be a good tool for these two groups?

And marketing, I think, gets a little bit more of a pass in times when it's directionally correct is okay, and, like, mistakes per 100 statements probably could be allowable as a higher number than it could be- Yeah

in, in a legal or finance or technical support s- side thing. So it's just kind of... I don't know. I it's a lot of the reaction that I've been getting, but also it's one that I'm struggling with right now because the capacity question, getting back to, circling back to that point, capacity questions don't work with this if it's not as good as a colleague.

Yeah. You can't say that this one saves you 40 hours a week, therefore I can use it like I would use a colleague because of the exact things first that Ruthi brought up, which is exactly it, that human beings are far more capable than a random task, and also tasks change constantly. So it's a really bad idea to try to automate the heck out of absolutely everything when things change.

I mean, everything's always fluid. It's not static, and that's, that static math that we sometimes use for annual planning or even further out-

...

Alex Pokorny: Is so bad, and we know it, but we still do it that way. Everything is dynamic, and they all react to each other, so you know, stuff happens. Yeah. Keep it fluid.

Dave Dougherty: But that's what I remember when I first-

Alex Pokorny: we don't really allow that. I don't know...

Dave Dougherty: when I first thought about pricing and learned about pricing in school, right? Like, I thought, it was like, okay, there's going to be some magic formula on how you come up with, you know, the margins or what the actual price is or whatever, and then you sit and you listen.

They're like, "No, we took the last thing, and we want X amount of things on top, and there's our number." And you're like, "So you're guessing?" You're a professional guesser. Okay. Cool. But also, wow, is that disappointing. Like, I I thought you knew what you were doing.

Alex Pokorny: Dave, the more you look at basically anything with economics, the more you realize- I-

it's a whole bunch of people just trying their hardest, and it's not right.

Dave Dougherty: We're all just weathermen.

Alex Pokorny: And they know. Yeah. Exactly. Forever weathermen. Just taking a guess, man. It might rain, it might not. I don't know. With

Dave Dougherty: periods of light and dark, intermittent. Anyway, Ruthi- I think that's it... you're getting frustrated with us.

What would you like to add?

Ruthi Corcoran: Oh, I don't know. I'm dying in a little bit inside. I could talk about price corridors, and the role of the market in learning prices, and, like, how the whole, the whole engine works and how we learn from... But we're- Mm-hmm... not going to, we're not going to go there today. We... Check out economics- There's definitely-

of pricing and the information theory of prices in your spare time.

Dave Dougherty: Mm-hmm. Well, actually, this is, this is the perfect kind of tie-in, right? There's the way things should work theoretically. Like, new tool rollout will be great. People will use it, happy days. Reality People don't pay attention to the training.

They're on their phone. They don't want to learn it, but they're there anyway because it's mandated training. You get, a percentage of the benefit that you thought you were going to get, right? Because you're dealing with people. People are messy, right? And I think it's, it's sim- a similar thing with, um, you know, we talked about AI for creativity and whatever else.

I think the fundamental problem with AI for creative, uh, use cases is that creativity is not a problem to be solved It's a state of play, right? Like-

Ruthi Corcoran: Same

Dave Dougherty: with

Ruthi Corcoran: prices.

Dave Dougherty: Ideally.

Yeah. But I think that, yeah, those are, those are some fundamental issues at least that I've been, I've been, wrestling lately.

Ruthi Corcoran: I think that statement you just made though, Dave, it cuts through the heart of a lot of it, which is when we're, we're thinking about human systems, human interactions, or maybe just any, maybe it's not just particular to humans.

Maybe this is just, this is how life and reality works. It's ne- it's never just a problem to be solved.

It's a, there- there's a process. There's feedback mechanisms. You know what Alex said things change. It's we don't live in a static world, we live in a dynamic world that- Mm-hmm

that is continuously evolving, and that's just as true when you're working on creative endeavors as you are with any, any number of endeavors that we attempt to take on

Dave Dougherty: All right, Alex, you have a second topic, so

Alex Pokorny: Oh, sure. Let's hit it.

Creativity vs Average AI

Alex Pokorny: So the second topic is basically just we're going to t- hit a little bit of the creativity aspect of it as well, of, you know, is AI kind of giving you the average of the internet, basically, and is it giving you the expertise or is it giving you the average?

And it came up more in depth, uh, Hank Green, if you know him from YouTube or any of his very many projects that he's worked on kind of had a long expose of basically the AI usage that he's had within his own creative process when producing YouTube videos, and how he's looking at reducing it. 'Cause he has some concerns of it, and how much it's been influencing larger amounts of production, which is something that he always feels pressured as a content creator to do.

Right. But on the other hand, it's... he st- is starting to think that it affects his quality. And some of the commentary kind of around it is also is he's a very, uh, scientific-minded individual, so he does a lot of research, and using AI for research means that he can do tons of information really fast, and he called, like, the dopamine hits coming from using AI all the time has been fantastic.

Right. Um, the problem on the other side of it is that he's getting that average opinion about all those topics versus him going through the struggle and the process, and then reading other people's information, and coming up with his own bias. So thinking of bias as a, as a positive thing in this way, because he's not just repeating the Wikipedia article, the encyclopedia, you know, dictionary definition of something.

Mm-hmm. Instead it's his take and his view on something, which is actually the thing that people want to watch and think that, they find interesting. I think it's the same thing with, like, really good professors. I had a terrible teacher once, he was a math teacher, who sat at the front of the class and literally just read the chapter to us.

Ruthi Corcoran: Oof.

Alex Pokorny: As you can imagine, grades were poor in that class. Later got fired. There was an assistant principal who came in to teach the last couple of days of the class. It was, it was a mess. They had a final exam of, I think it was five questions, and we spent two weeks on those five questions, and as long as you got any of the questions right, you passed.

, It was a mess. Um,

Dave Dougherty: but- You know, I think I could pass a test with only 20%. At that time Google.

Alex Pokorny: It's like please guess the right thing." Our school's numbers can't handle this. Also, best of luck on the next, next level, 'cause you don't know anything. Yeah... but just that, that, you know, really good teacher isn't one who, speaking back, uh, the idea of narratives and stories, but also kind of has their own bias, their own story, their own narrative, that basically they take the series of events or- topic.

They put it in a way where it captures your attention. They deliver it in a way where it captures your attention. You know, it really pulls true.

The other side of it is, you know, g- giving people a stack of papers, right? And going through this discovery process not using AI, you're going to come against a lot of different people's bias and their narratives, and it helps you kind of come up with your own, the ones you trust, the ones you push to the side, the ones you've eventually decided to push to the side, ones you've really nailed on or questions that kind of came up in your mind as you're going through it and learning about it.

That is that creativity which creates that content, which is different than what AI gives you. And that's the piece that I think, I mean, we've, we've talked about a few times in the past a bit about, you know, AI outputs being, you know, not the same as a person's bias and inputs and creativity and their, and the trust that we have for them, and then therefore we trust what they say when they tell us something-

Ruthi Corcoran: Right

Alex Pokorny: versus the Wikipedia article. Well- I don't know, what are you guys' thoughts on that?

Ruthi Corcoran: That's such a cool line of thinking, and like, uh, something to turn over and play around, 'cause I, I like this phrasing of like is it just the average of the internet? And then my, my thoughts immediately go to like, well, average can be good for like lots of things.

There's a lot of things for which you don't need the extra time and attention. So for example, me and cooking. Claude and I have been spending a lot of hours in the kitchen. It's been fantastic. I think I've been cooking not just more, but more effectively than I think I ever have in my life now because of Claude.

Like this weekend, I took squash blossoms, I stuffed them with ricotta, and I fried them into this lovely batter, and they were delicious. This is something I would l- never have tried with just the recipe that I found online because I, I need the extra back and forth, the extra content. But average is good enough.

Like, I just need the average way in which you heat up sunflower oil, cover the thing in flour and cor- like I j- I don't need anything in depth or some particular person's view of that because I'm, I'm just making dinner. And there's like a lot of use cases where that sort of average is good enough for what I'm trying to do, but then there's this other piece of like- Bias and discernment are also super important, but also super high cost.

Like, it's much more ti- time intensive, energy intensive to figure out, like, what do I think? How does it relate to what other people are thinking? And I think there that it's totally right. I can't just take what Gemini says and be like, "Oh, that's my opinion." There are certain cases where bias and discernment matter a lot, and I think a, in large part it's what matters to you or where is your sort of specific differentiation.

So in my job, I can't just repeat what Copilot is telling me, or I don't think I would be very good if I did that. Similarly, with my, my gardening, I'm not just going to be like, "Claude, give me a garden plan. I'm going to plant it." 'Cause I so enjoy the creative process. There's other people who are just like, "Just tell me what to put in the ground and I'll move on with my life."

And so I think there's a place for both, and it's what are the trade-offs we have to think about between, hey, this is good enough, versus this is where I think me personally I'm going to spend extra

Dave Dougherty: I keep coming back to this thing with mu- music production, right? Like, starting out with an acoustic instrument, right? The guitar, and having to go into a studio, and saving up your money to be able to book out time in a studio before all the home recording was a big thing, right? Like, I remember mapping out in, like, seventh grade how many months it would take me to save up enough money to have enough time for a day in the studio, and how do you actually work efficiently enough to record as much as possible, in that day.

Whereas now, having some conversations, it's like yeah, we have some new packaging. Can we just Photoshop that instead? 'Cause I'd rather sell the product." You're like, "Crap in, crap out, man." No, you can't just digitize it, when you're in the studio, right, you can do a lot of these electronic fixes with whatever you record, but if it's a r- a crappy recording of an instrument, there's only so much you can do to make it sound better.

You're going to have to just sit and redo the take until you get a good take, right? But then you have entire genres like electronic dance, right? EDM. It's its own thing. Would you say it has a lot of soul to it? No, but that's not its purpose, right? The purpose is to get you jumping and having a good time and forgetting that you hate your job, and you don't really like your boyfriend, and, you know, you're having a good time in the middle of the field with a lot of loud music.

That's great. That's wonderful. And who doesn't love a good light show, right? But it is so on time, it is so quantized that push/pull that you get with, sort of organic music it doesn't have the same kind of personality, right? One of the, the perfect examples of this is actually The Rolling Stones with Satisfaction, if you really want to nerd out, Keith Richards' guitar is massively out of tune And they recorded it anyway. Why? 'Cause they were high as kites. But that was part of their process.

Ruthi Corcoran: But I think you illustrated such a good point of, like, where's the value add and where does it matter? Like, where are there parts that are good enough is fine- Mm-hmm... and where are there parts where we're like, "No, this is actually going to result in the end product." There, I could see products where, yeah, just doing a quick digital update or just, like, Photoshopping it- it's probably good enough. Like, move on, 'cause that's not the main va- like, the product image may not be the main value add in that case. It's- Right... somewhere else in the process. Right. And, and so that's what's, that's what's cool is we have a whole suite of tools.

Dave Dougherty: You know, with

Alex Pokorny: the- Yeah, we'll defend the GM for Alistar For at least our European listeners, if not our American listeners.

There's good EDM and there's bad, and I think you're hitting- Of course... the nail on the head between some of that same kind of piece of some of the AI created music I think actually kind of hits a, a similar note, even if it is acoustic or hard rock or, you know, whatever genre, where if it doesn't have that element of, I don't know if it's expertise or soul or creativity or something to it, it's very bland and it's not interesting.

Mm-hmm. And yeah, it may be good enough to be a one track between many in like a long DJ set, but it's really not going to be the highlight. There's got to be, like, there, there's, there was- there's definitely ones who, you know the individual behind that had skill in what they were doing in creating the momentum, the, you know, the kind of the climax of the song of all those different elements to it to kind of like really build it into something.

But yeah, Ruthi, I'm kind of, I'm kind of coming to the same thing of categorization of like there's creative times, there's utilitarian expertise, kind of value add. I don't know. I'm trying to like... There are really good times when AI makes sense, and there's really times when AI doesn't make sense. Like your point about like planting, like there's a part of my lawn which honestly if anything would grow there, that would be great.

And if Claude just told me like the exact plan, I'd go buy it and I'd set it down and that'd be it. However, there's also when my wife and I went to like a garden center and we were thinking about a particular garden, we had a lot of fun with just like coming up with a random budget number, and like s- grabbing pots and stuff and just like literally taking over some of the sidewalk of the, the garden center and just kind of like laying it out and being like

And it was super fun, and that's basically what we, we planted and that's kind of like an element of that creative approach too. So there's... In that same piece, same vein, one of it was very utilitarian, but the other one was, I'm saying, creative, and I would totally AI one of them. But I absolutely would not, you know, replace that experience that I had with my wife at that garden center with AI either.

Dave Dougherty: Well, and specifically about like food and music too, right? I mean, it's so subjective to the individual, right? Think about- Yeah, that's true... all of the arguments about Italian food and like, oh my God, if you do this, the grandmas are going to roll in her grave and her ghost will forever- Yes

haunt you because cracked the spaghetti before you put it in the water. Or God forbid you're the type of person that does cottage cheese in your lasagna. Like just I don't need to know you- Right... if you are that. You know, but like those are the kinds of things that make it fun.

Ruthi is

Alex Pokorny: just dying on mute.

Ruthi Corcoran: I don't understand what's wrong. I do both of those things. It's great.

Dave Dougherty: You know, I mean, again, we can't all be perfect, and we can't really seek perfection like we said in the first segment.

But I think, like, for those things, like, those cultural nuances and those, those arguments, right? Who's better at cooking, Italians or French? All you got to do is say that in, in front of two of them, and then they'll just fight for the rest of the time, and you have your entertainment for a while.

That's fun. I love having those debates. Yeah.

Ruthi Corcoran: And the answer

Dave Dougherty: is Japanese. You're not...

Yeah. Yeah. Any kind of craft, that answer holds up.

Ruthi Corcoran: Okay.

Dave Dougherty: Yeah, for sure. But yeah, I think it, it is, it is interesting, and it goes back to, you know, a lot of the other conversations we've had about where's the appropriate use for these things, and how do you figure it out, because there is going to be some messy middle stuff that just is never going to be figured out, and y- I've been thinking- And you go-... about this too with certain jobs, right? Like financial planners. I actually think... I, I have kind of a contrarian view. I think there are going to be more of them because people will go to AI to handle some of the simple questions. But then if you're actually going to invest your dollars, you're going to want to feel good about having it be protected with another person, right?

Or have an assurance with another person. Mm-hmm. I mean, that was the whole reason banks were a big thing because you actually knew, all right, Jimmy's got my money. Like-

Ruthi Corcoran: Mm-hmm...

Dave Dougherty: great.

Ruthi Corcoran: I think one of the, one of the takeaways I've got from this back and forth and dialogue is, like, there's a time and a place for where you want to buy some discernment and that higher cost, and then there's also a time and a place where, where sort of the average is good enough.

And I, I think a key piece is, like, upping the quality of good enough is still massively impactful.

Sure. Like, if everybody's good enough just increases a bit, and all of a sudden you- you've raised the level of everybody's w- productivity, output, y- insert the word that you want. Mm-hmm.

And that's, that's pretty powerful, and perhaps even makes more space for the higher time and attention towards the areas where we want bias and-

Alex Pokorny: Yeah, I think of that like YouTube tutorials. I think of my parents' generation of trying to do DIY things around the house, and-

Ruthi Corcoran: Yes...

Alex Pokorny: you look at some of the outputs, 20 especially after it's aged, 15, 20 years, and you're wondering, "What were you thinking?" Like, my gosh. But at the same time, if the level of education you have is what you randomly have been taught, and then maybe a phone call with a friend or a relative who might know a little bit more than you, and you're just going to go at it, yeah, that kind of is the output.

Like, the... Also, the expectations are kind of set to that. But then again, if someone gives me a 15-part YouTube tutorial that tells me literally every step of the way on how to do something, it's probably going to end up pretty much looking like that. Not- probably not as well, but it'll look something like that at least, right?

Like, that, that good enough is just, like, shot up so much with that. Yeah, I, I guess I'm kind of with... I was doing some, uh, vacation planning with Claude recently, and it, it, it is really nice for the beginning stages of it, at least, of- Mm-hmm... here's a ton of different options. You know, it just kind of speeds up some of that research process, which I appreciate.

Like, I go nuts into that kind of stuff myself, and I know that. So this also is a nice way to kind of give me a little bit of a, a starting point that cut some of that time that I'd rather not spend anymore on that kind of a pursuit,

Dave Dougherty: interesting.

Wrap Up and Callouts

Dave Dougherty: This has been a fun conversation. If you've made it this far through this, thanks for sticking around.

We greatly appreciate you. Uh, like, subscribe, share, leave a comment. Are you the type of person that uses ricotta in your lasagna?

Ruthi Corcoran: Do

Alex Pokorny: you burn

Dave Dougherty: the house? Uh, or are you causing earthquakes in Italy with all the grandmas rolling around in their graves? Um-... let us know.

Ruthi Corcoran: High

Dave Dougherty: protein, man. Let us know. I- Cottage cheese.

Ruthi Corcoran: Oh, man. We

Dave Dougherty: made too much of this. What can we do with it? Um, anyway, anyway, take care. We'll see you in the next episode two weeks from now.