How Organizations Can Thrive in the Human + AI Era with David Chestnut

Speaker A: Welcome to the Edge of Work podcast. I'm your host, Al D. This is a podcast for leaders who want to make sense of workplace trends and are looking for new ideas about how to lead people and grow their business in a changing world of work. During each episode, I'll bring you the latest experts, researchers, founders and leaders to share new and unique ideas, as well as actionable advice around attracting and retaining talent, developing people, and building healthy and sustainable organizations.

Speaker B: Welcome to the Edge of Work. Today's guest is David Chestnut, who is a principal director of Human plus AI Talent Strategy at Accenture. David, it's great to have you on the Edge of Podcast. Really enjoyed our conversations. It's nice to finally hit the record button. So, just to kick off the conversation, would you mind sharing a little bit more about what is your role and how would you describe the work that you do?

Speaker C: Absolutely.

Speaker B: It's good to see.

Speaker C: Al yeah, like I said, this has been a long time coming, so thanks for letting me kick through the door here. So, yeah, Human plus AI Talent Strategy. It's a joke kind of title. If you were to ask somebody five years ago, they'd be like, what the heck does that even mean? Now it's the second fastest growing job in the world. And if you think about it in those terms, it's all three parts of that. I help human beings work better with these new tools that have kicked and elbowed their way into every single part of our lives. I design AI tools that work better with how human beings actually work. And I really design systems of work and processes and do the boxes and arrows work of drawing a big line on the board. Humans up top, machines underneath. Humans provide the expertise, AI provides the effort, and we design systems around how those things work best together. So, as you can guess, I'm a busy guy these days.

Speaker B: Busy indeed. I think I want to jump right to it in something that you said there. I know that you've been a technologist at heart for quite some time, and so this, the technology that you've been working with people on is not something that's new. But certainly I know that some of these, some of the things you're doing these days are really spending time with individuals who are trying to perhaps work with AI in some kind of capacity. But maybe they know about it, maybe they know they need to be using it, but maybe they aren't quite sure the best ways to go about it. So perhaps rationally it makes sense for them to be using it, but maybe they just haven't built that muscle memory quite yet, I guess. Just to start this conversation, as you've worked with individuals or leaders around adopting AI, any insights or takeaways from from working with these individuals? And I'm particularly keen. Again, technology adoption is not a new topic for you, but I'm just curious if there's a delta at all between what you've seen people adopt in the past versus what we're seeing with some of the tools that we have exposure and access to today.

Speaker C: Yeah, absolutely. It's funny, I, uh, think one of the biggest traps right now is to think of this as a technology adoption problem. And it really is fundamentally different in every single way. Because first of all, I don't lay awake at night worrying if Salesforce or Kubernetes is going to take my job. I worry existentially about this. And every single person has an opinion on AI whether we want them to or not. Some are positive, some are negative. So you're already walking in the door when someone says, my boss says, my company says. I say I want to use AI more. You're walking into someone who has a loaded perception of what these tools are off the Rick. And you and I both know it is an absolute hornet's nest of misinformation in the world. It's 50% advertisement, 45% hype and 5% real insights. And filtering through all of the garbage that AI, uh, has made it, or made it easier to create is one of the hardest things right now. So the insights I get is I first really think people need audit. So learning guy, right? I really think people need to fundamentally understand exactly what LLM based AI is. And we start there. And it is a. When I teach a 101 version, the first question we ask is when I ask AI a question, where does the answer come from? That's fundamentally an important question. And we assume that intelligence is that last word. I always think it's artificial effort, maybe artificial work, but intelligence, hardly. And so I think getting people in understanding that, hey, this is just bias, bottled cognitive effort. It's just something you can sprinkle on a task and get a reasonable output. How you inform it, what you point it at, and how you inspect what comes out is still 900% up to you. And so this is just extra effort that you don't have to spend to make an output. So the real resistance I think comes around from AI is just anything a computer does that I don't understand, which is what a lot of people won't fit into. And so we have to Say no, we're talking about LLM based AIs, we're talking about large language models. Next token guessing machines that are then given a scratch pad for chain of thought reasoning and then given a whole bunch of data that hopefully it remembers. Hopefully it doesn't start there. And then you can take people through with like, all right, we jump into use cases, we jump into other things and it just, it's too much too fast for most people. So I typically start these engagements with stepping back, getting a common set of definitions and then building forward with what is this thing actually capable of doing, what's it good at, what's it bad at? And then build from there.

Speaker B: I appreciate that line of thinking. I think you're right in terms of recognizing and believing that it's not really an adoption thing and it actually. Not that we need to relitigate the past, but it makes me reflect upon, I was going to say question, but I'm going to be a little bit more generous but reflect upon maybe the way in which we've approached how we use technology to prove the way we work fundamentally over the past 20, 30, 40 years. Maybe that's actually the culprit part of the culprit of why we get to where we are. When uh, what I'm hearing you describe to me sounds like things that we probably should have always been doing.

Speaker C: Right. A hundred percent, I would say when I talk about this being fundamentally different. I had one of those shower thoughts a week or so ago and it was one of these human beings always did the work. We created the things. Yes, the systems carried the context. Right. They were the ones that the system of record, the CRM, the lms, whatever it may be, it carried the context for us that has flipped. AI is now performing the work. It's up to us to carry the cognitive load, the weight and the context between tools. And so that's a very big ask of someone whose whole job is to m make outputs their whole life. Now outputs are easy and free. Everybody can make a B minus with no effort. Right. If I know how to type a prompt, I can get them be a pretty good output. So we've now raised the floor of human capability. I don't think we've done enough about the ceiling just yet. And I think that's where a lot of organizations are stuck. Sure, yeah, yeah.

Speaker B: And I appreciate even what you just said about the focus on the output. Right. And if I were to describe knowledge work in, I don't know, we'll just say three words. It is decks, it's deliverables and it's documents, right. It's output. Right. And let's be clear, output matters. But I think part of what has been so interesting to me, and I know you've worked one on one with people before, is when you see someone have one of those aha moments where they, they see the output come like that, it does become that next level question. If they are reflective about it of oh shit, well then what do I do? Right. The caveat of we didn't check it, we didn't make sure it was, it was good and all that. But it does, it does make you wonder when you think about what you've been focused on in your role or your craft, whatever it is for all these years and the effort you put into that document, that deliverable, that deck, that etc, when it can just go

Speaker C: like that, you start to say, oh,

Speaker B: what do I do now?

Speaker C: Yeah, it's, it's tough because again I talk to technologists every single day. Code being one of the biggest things that.

Speaker B: Sure, yeah.

Speaker C: Uh, these tools at right away, right, Knew they were probabilistic tools. Their limitation was declarative capability or so they all right, let's teach them how to build something that's deterministic, not probabilistic. And so we build code. But I talk to people, they write code like art. And I think we're in this rough place now where we are being asked in knowledge work and in a uh, lot of creative efforts with AI to disaggregate the art from the craft. And that's a really big ask for somebody who spent their whole life loving the craft and the art was the byproduct. I like the journey. I like, I don't just want food. And I think that disaggregation has been a lot, uh, really hard for experienced, expertise ridden folks like you, like me, who have years and years of doing something one way and becoming great at it, being told first of all what you're doing is the wrong way now. Second of all, you have to use this tool that we just showed up with. And third of all, there's a little disclaimer at the bottom that says this is probably crap, don't listen to it. That's a really rough world to ask people to live in.

Speaker B: It is, I want to take the other side of the people that you're describing, those experts as people with deep expertise, with lots of knowledge and wisdom that's accumulated over the years and talking about, talk about another population which is early career talent. And I know that you've had a chance to in particular, just speaking about a company like Accenture, they hire hundreds if not uh, thousands of early career talent every year. But I know that in particular you have spent quite some time with early career talent as it relates to how they are uh, working with AI and how you're working alongside them, working on using AI to collaborate with, work together. I am curious what have you seen here in just general observations and then maybe the two parter for this. What have you seen that maybe is missing from people who just read the headlines and just see that oh, early career talent is going away or uh, maybe who don't have the privilege or the experience of actually working directly with early career talent that they should know or that they should be aware of because they just haven't seen it.

Speaker C: Oh, I love it. Yeah, it was funny. I worked in L and D for a long time. For those who don't know my background, I ran Accenture's technology learning strategy for the last eight years or so. And so everything from super advanced to super brand new folks walking in the door. But the thing you don't ever get to do in L and D in the corporate world is teach. And you talk about that moment of the lights coming on in someone's eyes. And so one of the things when I moved back into client facing consulting work was I said I really want to teach again. And so I've been able to for the last year teach every incoming group of consulting analysts that have come into North America and Accenture. Around about a thousand folks and you're seeing a couple of really cool things emerge. Number one is they now get a full day of AI and data. They get two or three days of an application of AI and data technology. Not just tool training, but delegation, discipline, balancing capability and capacity of their AI teams. Token efficiency. Thinking about uh, things other than just how do I get as quick as possible to make the deck so that I can go back to doing nothing. We've tried to stop framing AI as an efficiency tool, right? You and I talk about Jevons Paradox. Like we all understand efficiency doesn't make you use less of something, right? And so what I think the big mandate that the earth shattering headlines are we're not going to need early career talent anymore because we're going to automate all their work.

Speaker A: Cool.

Speaker C: Who becomes the mid career talent in five years? Who becomes the senior talent in 10 years? You don't. If you believe my premise that AI is just bottled cognitive effort, CEOs, the rest of the company goes cool. Effort equals outputs. Outputs are what we pay for. Use the AI to make the output. We can all retire early. But as learning guys and human capability folks, we understand AI is also cognitive. Effort is also critical in the development of capability. And it used to come along for the ride for free, right? You built the thing, you started out pretty bad at it, then you got okay at it. And maybe you became good at it, eventually become great at it. But all that happened through effort, reps and sets, doing it wrong, being bad at it for a little while. I think there is a tendency in companies to rob early career talent of those skin knees by going, just use AI. I think the second part of that is what do you lose by doing that? Besides the obvious, you don't have experts anymore, fine, that's Ford that needs to get fixed. That's a problem that's going to manifest. That's a, that's. Unfortunately it's a pretty long fuse on that. I think it's longer than the tenure of the average CEO. And that's the scary part, right? You got a three year average tenure of a CEO in corporate America, you get a five year fuse on this talent evaporating. Roughly. What I think the mandate becomes for companies is I started thinking of the L and D almost as effort architecture. We have tools that can take someone's effort almost down to zero. Our job as L and D professionals is to, yeah, give them the training, do the things up front. But all these tools that show up in the flow work, decide how much actual intentional effort needs to be left in the task to make it meaningful for building capability. And so instead of giving somebody access to CLAUDE code and just going use this to write code, build them an agent whose job is to go, hey, your job here is to write some code. My job here is to help you, but I can't give you the answer. Or you can write a PowerPoint, I can build the deck for you. You have to decide what goes on each slide. Those things that become inherent to the 10,000 hours, the reps and sets that make you great at things. So leaving intentional effort, good toil, right? Good effort in the system at a diminishing degree as someone gets better. I think is where we're going to see things go in the future. And we're trying things like that as we work with early careers. But it's hard because I have people walking in the door that sit down at Claude code or sit down at cowork or sit down at any other tool. Take your pick. And they. I call them a natural stick pilot. They've been waiting for this cockpit their entire lives. The problem is they have infinite build capability and they do not have not built the expertise or the taste to determine what's worth building. And I think that disconnection is creating more problems than we realize in the. In, in the near future, because again, I can build anything, but I don't know what's worth building. I, uh, I wanted to ask a

Speaker B: question to tie a couple things together. So you mentioned earlier around this idea of being able to use AI to erase the floor versus using AI to elevate the ceiling. Right. And for those who aren't familiar with that analogy, it's are you bringing maybe the lowest common denominator up, or are you actually unlocking to a level that you could not have seen previously? So that's one train of thought, and I think that's something I think about a lot with early career talent in terms of could you find a way to take an incredibly capable set of individuals and create something that allows them to unlock all the capabilities that they have, all the things that in many cases you're paying them, and particularly, no offense in the case of Accenture, a pretty penny to be there for, and so that they can actually put that insight into use. But I'm trying to square that with this other thought of what you said, where you do get people who right out of the gate, they can be that natural stick pilot, right, where they can build anything, but they haven't yet necessarily gotten the sets and reps to know what they should be building. And so I guess the question that I'm trying to figure out, and I've been thinking about, at least is there a way or have you tried anything in the idea, uh, field of accelerating that sets and reps, accelerating that, the judgment or the taste or whatever you want to call it, so that it kind of marries the two of those things of being able to raise the ceiling, but doing so in a way that isn't just about building a bunch of stuff, but actually building that judgment and taste.

Speaker C: Have we done it at scale? Have we done it as much as I'd like? No, because I think it's becoming a newer tar.

Speaker B: Was. Yeah.

Speaker C: Yeah. Well, when we were at. Even when we were at asugs, uh, we had folks from Lovable there, and they walked in the door. Everyone's a builder now. I don't want to be a builder. I hate writing stuff.

Speaker B: Yeah.

Speaker C: I build stuff. To tell stories, because that's what I love doing. I love communicating. So what we've started doing is, I'll give you three words, pairs, pods and solo pilots. The idea of. It's funny with human capital management or whatever you want to call our discipline, people, nomics. If essentially the old craft is becoming cool again, the new stuff's becoming cool again. And I think one of the biggest ones is team dynamics, right? Which team is which people work best together? What are they drawn towards? What are they capable of doing? How do those things flow together in an organization? We used to optimize building teams around the scarce resource, which was the ability to build stuff. Now we can optimize teams around anything because the building is, to everyone's point, is now no longer the scarce thing. So we take those new builders that walk in the door with all this capability and no taste. I have other archetypes that I think exist in work, and I think one of them is builder, one of them is translator people like me that love to tell stories and contextualize things for the audience and think about the audience. Then I think you've got arbiters, people who know what's worth building. People would have that late career judgment, but maybe they never want to be a builder. You've got framers, people that know how to take problems and turn them into suggested solutions. And then you've got. If you just had those four, they would just chase each other in a circle and never build anything. Then you take conductors, people whose whole job is to move the process forward. And I think if you work in those constraints or if you think about in those capacities, there's a couple ways to get that. Most of the time you have a senior person who is not AI enabled, you pair them with a builder. There's your pair, the senior, the junior. They learn from one another. They built cool things that can never get built before. You've got senior folks that learn AI that are enabled, folks like me and you, who have the expertise, the discernment. We have enough of the other archetypes. We can build anything we want now with AI, we can push our capability forward, which I feel is what I'm doing at work. I feel like a lot of folks are doing. And then you've got pairs or your pods, right? You just take one of each or the number of each you need, and you put those into a work group. And we're trying to test that out in some very limited spaces. But I wouldn't say that it's pervasive

Speaker B: Yet I think the net of it and what I'm hearing of what you're trying to do is that, forgive the simplicity of the statement, but it takes work and. Yeah, and it takes effort in a way in which I don't think that we collectively, particularly for this population, have really invested time and effort and energy into in the past. Perhaps even really if we look at some adjacent professions, if we look at the trades or we look at professions that have a licensure where this kind of apprenticeship and this kind of onboarding is literally baked into just the certification or the licensure or just the way in which a new plumber or a, you know, a new electrician or whatever it is ends up becoming a more senior one. And while it does feel like the nature of the ethos of doing that is, is commonplace in those fields, medical is another one. It's not something collectively, at least in this knowledge work world that maybe we have valued, I would say outside of professional services, broadly speaking, as a category which I think if anyone's done it or done it better, I would say it's them. And it's because I think they, they recognize, or they're at least recognized in the past that if they have better people with more capabilities, they can make more money.

Speaker C: Yeah, we sell the time and attention of smart people. That's the way I always define what professional services does. And you've got to create two things, then you have to create people with time and you have to create smart people. Right? Yeah, it's funny. 70, 2010 and all of these blooms and the two sigma problem and all these learning science things are again are rushing back into the forefront because we now have these fun little stochastic parrots that they can do work for us. We have to now sit back and go, wow, how do people actually learn? My favorite thing I think about AI in recent years is just how much it has taught us or how much helped us re remember or rediscover so many things about humanity. Everything from the, the far end of the hard problem of consciousness all the way back down into oh yeah, people really do learn skills by getting a little instruction. 10%, being around other people that are good at it, 20% but then doing it in a safe to fail environment is 70%. We tried to shove all that. 70, 2010 led to training classes and have people sit there for three weeks and then the forgetting curve kicked in. But then what consulting did was send um, you to training for a week, put you on a project with other smart people and then let you do it for six months. Guess what? That's how capability. So it was 70, 20, 10 built into the operating model. And I think as cost pressures have come, as other things have come, you can't just back up the truck and have 40 analysts pile out with the two seniors and run a consulting project anymore. You now need every single person has to be contributing at a higher volume, creating more outputs. And so consulting and other disciplines have over rotated so hard to the efficient delivery of effort. And if you believe anything I've said so far, effort is on a collision course with free, or at least free enough to be a rounding error. It's important for us now as consulting folks, as knowledge workers, as human beings, to really sit back and look at what does AI do without us. And it's funny, I was talking to somebody about this the other day. I think if you threw a rock on LinkedIn, you would hear the word judgment in every single list of skills that's ever out there. And I'm like, judgment's not the kind of thing you hire for. You have to build it. Because judgment to me is just expertise applied. Right, it is. I know what matters and I'm applying it to a set of conditions. Judgment gets built into the. Under the conditions, judgment has to be your customers, your processes, your unwritten rules, your, your policies, and then the policies that are okay to bend, the ones that are okay to break, and the ones that you must follow. All of those things define a culture. They also define how judgment looks in an organization. Assuming you can hire for judgment like it's in somebody's backpack and it worked at their other company, so they have judgment at this company just doesn't translate. So I think the things that everybody talks about is the thing AI requires taste test, judgment. Whatever those couple of words are, they only get built in house. You can't bring them in from the outside. And I think that is really, really tough, especially in a professional services firm that's always under cost pressures. And at this point, they all are.

Speaker B: Hi everyone.

Speaker A: Al D. Here. Thanks so much for listening to the Edge of Work podcast. I hope you're enjoying today's conversation. In addition to hosting the show, I spend my time advising, coaching and partnering with leaders and organizations who are navigating a, uh, rapidly changing workplace. If you or your organization are, uh, focused on helping your employees and leaders lead through change, strengthening their human skills in the increasingly technology driven world, build greater adaptability, or navigate the AI change management challenges, I'd love to connect. Whether you're looking for a keynote speaker for an upcoming leadership event, a leadership program, or just want more hands on support through workshops or coaching.

Speaker B: I'd be excited to learn about you

Speaker A: and your goals and explore how I can help. You can find my contact information in the show notes. I'd love to hear from you. Now, let's get back to the show.

Speaker B: So I want to change topics for a second and shift more to a conversation. I think partially on the same vein, just talking more broadly about the intersection of AI and human beings for that matter. And to talk about this, I want to pull from some of the writing that you've done on your your Medium newsletter, which is around the authenticity revolution. And I want to pull a quote from this and I'll let you explain the concept of the article and then you can certainly talk a little about what this quote is, which is you write that quote, unquote, AI slop is dead along with human slop. And so what is this idea of human slop? I think a lot of people are familiar with work slop or AI slop or have experienced it themselves. But what is this concept of human slop? And I think you write about how it's going to become more valuable. And so I just would love you to unpack some of that for all of us. Yeah, absolutely. It was funny.

Speaker C: This whole article got inspired by literally a trip to a museum. And it sounds like a fake story, but it really happened. It's the museum in Chicago, it's Field Museum. And you're walking through and you've got all these antique, these clay pots from 5,000 years ago. And I walk into the room and there's a group of people standing around this one pot. In the end, they're all pretty much identical by my eyes. I get up close, one of them's got a thumbprint on it that the artist left there, some, you know, Etruscan era, um, however many thousands of years ago. And we all cared about that one. So I think when I talk about human slop, let's talk about the AI side of it. What was AI trained on? Where does its training data come from? It came from Reddit, it came from social media, it came from Wikipedia, sure, Shakespeare, but also 4chan. And so like the two extending ends of human history. And then what happens? We reinforcement training. Any system large enough is designed to cut off the sharp jagged edges, round itself out. So AI, I think, always optimizes its outputs. And then what happens? You get, you don't get peaks and valleys, you get A straight line. And that texture that we lose gets rounded off again by AI that we use to create outputs and rounded off again, and eventually everything looks the same. Have you been on LinkedIn lately? Have you been anywhere lately? But I won't go through and rattle off AI hallmarks because that's a little hacky at this point, because we all, like poor M. Dash, just gotten, um, beat up to no end. What's the false dichotomies? It's all those things, and it's a hallmark of that AI slop. And so if you believe again what I've said, that AI is an efficient replacement for effort, not expertise. You and I are experts, at least in one thing. We can decide what they are. How often do you optimize your outputs perfectly? You never do. You're hungry, you're sleepy, you're tired, your brain's not working that day. You drink too much last night. You didn't drink enough last night. Whatever it is, your outputs are always a little off. And that little off from the norm is human slop.

Speaker A: It is the.

Speaker C: It is what AI needs to create outputs. And what's interesting is, you know, the perfect example of this is model collapse, right? Let's try and train AI on AI generated data.

Speaker B: Guess what?

Speaker C: It's perfectly optimized. So you train a model on AI generated data. The ouroboros just eats its own tail. It is. It becomes gibberish. It does not work. And so it requires. I think about Back to the Future, too. That image comes into my head of they pull up, and instead of having to run on plutonium, now the DeLorean runs on just garbage out of the trash can. It's Mr. Fusion turns it into something amazing. And again, I call it human garbage. I call it human slope, partly for clickbaity effect, but also because. Let's just mean an unoptimized output. But guess what? Every Picasso is an unoptimized output. He tried his hardest. But we know artists suffer, they have pain. That's where art comes from. And I think it's the surprise of the unexpected that makes human beings love art so much. And I think that is re. There's a requirement here that if you create something that is by its nature rounding off the rough edges and the uniqueness and the interesting bits, that you're always going to get something that feels a little bit less. And I think human beings are detecting it on a cellular level as evolutionary competition. I think we're definitely connect, picking it up on a cognitive level because we are Inundated with it. And this pattern matching machine that we have in our, this pattern seeking computer that we all live inside of goes, that's AI. We don't even have to know anymore. You read it, I get a sentence in and I'm like, I'm not reading that. Because there's also this concept of effort, right? This idea of if you're not going to put in the effort to read it or to write it, I'm not going to put in the effort to read it. And I don't know where that comes from in me, but I see myself doing it. I talk to lots and lots of PPO people from CEOs to entry level careers. All of them are saying the same thing. If I read it and it sounds like AI, I'm just going to scroll past. So I think there's a precedent for this, right? The market always reprices for the real thing. If fast food becomes available, farm to table comes back fast, fashion vintage becomes popular. Even though you can say something built in a machine by a factory farm is the same amount of calories, ideally optimized, perfect, perfectly good. Then why do we pay $100 for a bruised tomato or whatever the thing may be? So I think there is a. Human beings are always seeking out to see us in the things we care about. We want to see the unoptimized outputs of us. It's an evolutionary thing, right? We're always looking for evidence of ourselves because it increases our chances of living long enough to make more people, which is essentially the two things all of us care about.

Speaker B: I also think, and I think you talk a little bit about this, the part of why I think the BS detector goes up a little bit more is just because when there is so much stuff that floods the market, when there's so much supply of things, of stuff, of content, of output, of decks, of whatever, you have to find a way to differentiate of some kind and to have a use your own judgment to, to figure out what, what is worth reviewing or what is worth evaluating, et cetera. And I think I'm um, looking at your piece again. You talk a little bit about this, where you talk about how when supply floods a market, the differentiator moves from access of it to trust. And I'm wondering how should leaders or for that matter creators think about trust in the context of work? When uh, AI does make that work cheaper, when it makes it what we traditionally think of work is it when it makes it cheaper, when it makes it faster, when it makes it more Abundant.

Speaker C: So I think about places where I wouldn't if I was reading this. Again, this is my translator kicking on. Right. I am always thinking about the audience and so content creator, show, let's say that's what I am like if that's part of one of the hats I wear. Right. I write stuff on LinkedIn, I create stuff on medium. I write that with my own stubby human fingers because I believe words are important. And, and I also think that now does AI help me in the writing of stuff. Yeah. I also don't put any LinkedIn comment or LinkedIn post or social media post that was written for me by AI. Number one, because of Hallmarks, because I think it's ugly. But number two is I'm asking you to stop your scroll. Look at this thing for five minutes. That's asking a lot. I can at least put five minutes of effort into it to ask you for that. And I think it comes through. I also feel again, my voice is unoptimized so much to the point where people have started putting intentional typos in things. That fad, I'm glad, has gone away a little bit. But you saw it for a while, like in the title of things Guys, how hard was it to misspell one word? We still didn't put any effort in. And so I, I think just the idea of authenticity becomes so much more critical for content creators to have something that is a unique voice. I think people will get over this sort of general, this generalized. If it's AI, it's bad. I think that'll come away when we start seeing utility from it and we start seeing it do things like. Things like I always. Everybody points at any time AI gets beat up. We all point at every cure, right? We go, look, disease stuff. It's keeping people alive. If you're not familiar with every cure, it's great. They're matching, using AI to match existing approved human molecules to see if it will work on other diseases. That's amazing. Great use of AI. It's not an LLM, let's be clear. But again, AI is painted with one giant broad brushstroke for most people right now. And so I think you get into this place where if you don't have enough realness in the world, the next version of the next frontier models are going to be terrible because everything it's consuming off of everywhere now is just going to be AI generated. So we will start to see societal model collapse. Because how much of social media and how much of content on the Internet is just regurgitated swap and it pops up in these little things and it pops up all the other places. But I do think you're going to see the market always reprices, right. It's a problem for limits. All the way back from the 70s the market's always going to say give me the real thing. We're going to see it in lmd. People want to learn from real experts. They don't want to learn from AI. I think they're okay to get rote information from a tool. They want to get stories from people, they want to get connection from people. They want to get emotional need from people. And I think that authenticity is going to just continue to go up. Goes back to our earlier point. If we're not creating any more authentic experts, we're going to be in a bad way in a shorter time than we think.

Speaker B: The following to that is if you can just be the expert because we're not creating it. There's your opportunity, right? And hard won earned insight from doing something and getting better at it. If you can be that, that when it everyone else is not doing that,

Speaker C: there's your alpha and you stand out and you. And again, I don't want to blow our own horns, throw my shoulder out patting us on the back. But the idea is that there's a reason that I think people with a unique voice and with real, genuine, authentic lived experience that can explain it to other people. Well and not just. I think we've had a lot of overnight experts in the uh, in the AI space. I think you again you could throw a rock on LinkedIn and hit 500 profiles that say future of work, you know, how. What does the future of work even mean to different people who became an expert in this yesterday or last week? I'm not trying to say that there's not good content out there, but when I read and it's all the same 27 words, just your version of Claude and my version of OpenAI and your version of Grok spitting it out in different combinations and permutations, it stops being original thought. And I think we're going to get to this point where the cycle just becomes so much noise that to your point filtering through the BS detector has to get better. But at the same time eventually there's just none of that left and it's all, you know, GEO reading, AEO reading SEO and we're just getting whatever falls out of the bucket shows up in our algorithms. And I think it's, I think you're seeing this Massive mass rejection of, um, AI. I think a part of that is. I think a part of it is evolutionary biology. It's something that human beings haven't felt ever, maybe, which is competition.

Speaker B: Yeah, I know that you think about and reflect on and study a lot of things, but if we take a time horizon, we'll call it the next recording this in the summer. So we'll just say in summer of 2026, what is maybe one or what are maybe one or two things that you're either excited about or exploring at this intersection of AI and work.

Speaker C: So I rattled off one earlier, which I think is this idea of these archetypes and these combinations of how do we get those sort of critical skills into groups that create work. I think one of the other things that's. I'll give you one that I'm excited about and one that's keeping me up at night. We'll start with one that's keeping me up at night. And I'm not going to go existential. I'm not going to go universal basic income. It's the end of intelligence as we know it. I think there's enough of that rhetoric out in the market. You can choose your doomsday prophecy of choice, go find the article that writes it and have your own sleepless night. As Ethan Malik would say. I think what I think about at night is companies are not designed for exceptional employees. They're not designed for exceptionalism. And so you take a tool like this, you put it in your hands or my hands or other people that I work with, and from what they were doing yesterday, from what to what they're doing today, they're 10x ing their output. Maybe 15, maybe 20. Now zoom out to the organization. Look at that individual. Look left and look right. They are 20 times more capable or creating 20 times more quality output than the person to their left and to their right. Can we compensate that fairly? Can we inspect that at that scale? Can we afford it? Hashtag tokenmaxing. So I think when they talk about the future of work and work needs to get redesigned, what we're really arguing and circling here is that work as we've known it doesn't exist anymore. With the outputs are readily abundant, easy to create, and completely capable of being produced almost instantly. So I think we have not hit the ceiling raising moment yet because most organizations are just going to design you out of it. Again, not, uh, to get too philosophical here, but you're going to get that rounding of edges that AI does. Any large system does that And a corporation is a perfect example of a large system. What happens to the top 20% of employees in a company? They either get promoted into a new job so they move back to the middle or they leave and go somewhere else so they get bad at it again, so they move back to the middle. Think about people that are behind the curve. What do we do with them? You either promote them to customer. Nice way of saying we fire them or we train them up m and get them back to the middle. So any large system is pushing things to the center. And I think if you take people who are truly exceptional, who have great domain expertise, have learned how to use the AI tools and can 5, 10, 20x their outputs, your system doesn't know what to do with that, so it shoves them back to the middle. You arrest their token spend, you tell them to stop. You take away their tools, you take the one thing they built and make that be their whole job for their forever, so they stop innovating. What excites me is that these people are out there at every level of every career and every job and there's going to be a handful of companies that figured out first what to do and how to deal with these exceptional individuals. I'm hoping they cordon them off, put them into a place and then put people around them who don't want to be that. They just want to take things and scale good ideas, make things happen. I consider myself in the camp of the former. As my wife likes to say. I'm, um, all jobs and no laws. I've got good ideas, I can tell stories. I have no discipline and no desire to go make it be real. And that for me, I think, is the thing I'm excited about, is how do we start harnessing these people who have always had the ability to be an artist, but they can now be art directors and give them this super intelligence or this increased effort machine, this cognitive partner, they can take what their ideas are, turn them into reality and turn them into something amazing quickly and efficiently in a way that scales to the world. We're going to see amazing things happen. I think once businesses get out of the idea of AI is about bringing up the floor that gets me excited.

Speaker B: That's, uh, a lot of great thoughts and a lot of great things to think about for the rest of the summer and beyond. David Chestnut, thanks so much for coming on the edgework podcast. If people want to read some of your work or follow up what you're up to, where can we point them towards?

Speaker C: Absolutely. I put everything on medium. None of it's paywalled. I think about this stuff all the time and every time somebody reads one I go, wow, really? But I hear all the time that they're somewhat helpful. So medium.com forward slash I think it's avid chestnut. I don't think it's hard to find. I'm sure it'll be the link in wherever outputs this and all the places that it shows up. And then, yeah, I'm on LinkedIn. Don't follow me, just add me. I'm sure there's something we can learn from each other, so there's no reason to be coy. I definitely want to know you, I promise. So find me on LinkedIn. Find me there. If you find me on Instagram, I hope you like pictures of cocktails and dogs. But yeah, that's where you can find me. And thanks for having me, Al. This was a blast.

Speaker A: Hi everyone.

Speaker B: Al D here.

Speaker A: Thank you so much for listening to

Speaker B: the Edge of Work podcast.

Speaker A: If you like what you heard, I encourage you to share the episode with

Speaker B: a friend, as well as to head over to Apple Podcasts to leave a

Speaker A: review and let us know what you think. I would be forever grateful if you did that. I would also love to hear directly

Speaker B: from you about what episodes you're listening to or any suggestions you have for

Speaker A: how we can make it better. You can find me on LinkedIn.