Episode 277 · 19 February 2026 · 54:12
What REALLY happens when you let A.I. run your workday, with The Economist's Boss Class Host, Andrew Palmer
Featuring Andrew Palmer
What REALLY happens when you let A.I. run your workday, with The Economist's Boss Class Host, Andrew Palmer
0:000:00
Tip: click any timestamp below to jump there and keep playing.
Enjoying the show? Subscribe for free
Chapters & timestamps
Click any timestamp to jump the player to that moment.
Intro: The AI sales coach that gets things wrong
Meet Andrew Palmer — Boss Class Season 3 goes hands-on with AI
interview#management#economist#podcast
Generative AI vs. the AI that came before
interview#generative-ai#ai-history#technology
What management tasks is AI actually taking over?
interview#management#ai-adoption#tasks
Hallucinations, mistakes, and learning to live with imperfection
What thriving looks like for humans in an AI world
Jobs are made of tasks — and some tasks are more automatable than others
interview#job-tasks#automation#work-design
Will AI save or destroy entry-level jobs?
Phase 1, Phase 2 — how organisations should actually roll out AI
Getting hands-on: how Andrew decided where to experiment
The unheralded hero: AI as a coaching and feedback tool
The standout takeaways from Season 3
interview#key-takeaways#season3#insights
What scared Andrew most — and the column that got critiqued
interview#ai-fears#critique#long-term
Be happily confused — advice for leaders who don't know how to feel
interview#leadership#uncertainty#advice
Advice for the nervous listener
Top 3 takeaways and where to find Andrew
Show notes
What REALLY happens when you let AI run your workday — with Andrew Palmer
Guest: Andrew Palmer — host of Boss Class from The Economist, author of the Bartleby management column, and Season 3's hands-on AI guinea pig.
AI isn't coming. It's here. And Andrew Palmer has spent an entire season of Boss Class living with it — testing it, breaking it, and asking the questions leaders need to answer before the technology transforms their organisations.
In this episode, Andrew joins Al and Leanne to share what he discovered when he let generative AI draft his management column, why the companies getting results with AI aren't the most tech-savvy — they're the best managed — and why human judgment matters more than ever in a world of hallucinating chatbots and AI sales coaches that get things wrong half the time.
No hype. No doom-mongering. Just a clear-eyed look at what AI is actually good at today, what it's still terrible at, and how leaders should think about deploying it in ways that help people — not replace them.
Top 3 Takeaways
The companies getting results with AI aren't the most tech-savvy — they're the best managed. AI adoption is a leadership challenge, not a technology challenge. The organisations that thrive aren't those waiting for perfect tech, but those integrating AI intelligently into workflows and decision-making.
Don't outsource your thinking. Andrew spoke to people at the frontier of AI who use it as a thinking partner — critiquing ideas, suggesting alternatives, helping make sense of complexity — but never as a replacement for their own judgment. The moment you hand over the thinking, you hand over the job.
Be happily confused. If you're a leader who's excited by AI but also uneasy about it, Andrew says you've got it exactly right. That tension is the correct response. It means you understand both the potential and the stakes — and you're paying attention.
More From Andrew Palmer
- Listen to Boss Class from The Economist — Season 3 explores AI, management and the future of work: economist.com/audio/podcasts/boss-class
- Search "Boss Class" on Apple Podcasts, Spotify, or wherever you get your podcasts
Connect With Us
Get in touch, book a call, or find Al and Leanne on LinkedIn — all in one place: truthliesandwork.com/connect
Mental health support: findahelpline.com — UK: Samaritans 116 123 · Mind 0300 123 3393 — US: 988 — Australia: Lifeline 13 11 14
Truth, Lies & Work is part of the HubSpot Podcast Network.
Full transcript
Expand full transcript
Al ElliottThere's an American tech company that's developed this AI sales coach, and it quite often gets things wrong. It misreads calls, or it criticizes strengths thinking they're weaknesses. It occasionally just misses the point. Yet even though it's not 100% perfect, it's driven a 50% jump in sales productivity.
Leanne ElliottThere's also a product manager who's built an AI version of herself, not to replace her judgment, but to deliver it faster. Her team gets better feedback, she gets fewer interruptions, and this AI isn't even close to perfect, but it's already making life much easier for her and her team.
Al ElliottOur guest today, Andrew Palmer, writes about management for The Economist and hosts Boss Class, currently in the top 5 management podcasts in the UK. He spent an entire season inside real companies like Johnson Johnson, watching AI actually land in the workplace, and what he found completely reframes the conversation.
Leanne ElliottThe companies getting results with AI aren't the most tech savvy. They're the best managed. And there's a clear message for the leaders still waiting for AI to be perfect before they touch it. Andrew thinks they're solving the wrong problem entirely.
Andrew PalmerEvery leader who's banging the drum for AI, who isn't, you know, playing around with AI, is not doing themselves a favour. You know, they need to develop their own intuition for where this might work, and where it's actually not going to be particularly productive. So, I think that would be my advice. So, be happily confused and experiment.
Al ElliottToday we're asking, what does smart AI adoption actually look like? And also, what separates the leaders who are thriving from the ones who are just frozen in fear? This is gonna change how you think about the whole conversation.
Leanne ElliottHello and welcome to Truth, Lies and Work, the award-winning podcast where behavioural science meets workplace culture, brought to you by the HubSpot Podcast Network, the audio destination for business professionals. My name is Leanne. I'm a Chartered Occupational Psychologist.
Al ElliottMy name is Al, and I'm a business owner, and we're here to help you simplify the science of work.
Leanne ElliottSo today we're back with Andrew Palmer, management columnist at The Economist and host of Boss Class, currently the number 1 management podcast in the UK. I say currently because we usually lose our position to Andrew when he releases a new series.
Al ElliottYeah, it's true.
Leanne ElliottSo yeah, and I did joke about this, Andrew, but we're coming, we're coming after him. But anyway, Andrew's 3rd season has just launched. It is brilliant, and it's entirely focused on AI in the workplace, not the theory, the stuff that's already happening in real companies with real managers right now.
Al ElliottAnd what Andrew found is genuinely surprising, and not because AI is doing this dramatic, terrifying things we keep reading about, but because the most effective uses are almost deliberately boring. And that's exactly why they're working.
Leanne ElliottAfter this quick break, we'll get into what AI is actually being used for in workplaces today, the 50% sales productivity jump from a tool that isn't even that clever, and the Hillary Gridley story that honestly might be the most practical management idea we've heard all year.
Al ElliottDon't go anywhere. The dirty little secret is that most businesses think they know their customers. They have the data, the records, the history, but it's scattered across 3 teams and 4 systems. And in our case, it's about 216,000 spreadsheets, which means nobody can actually use it.
Leanne ElliottWhat you should be doing is using HubSpot, because why? HubSpot connects it all. Every interaction, every support ticket, every conversation into one platform every team can work from. So when sales talks to a customer, marketing already knows the full story. And when you know more, you grow more.
Al ElliottNice. Check out hopspot.com, the agentic customer platform for growing businesses.
Leanne ElliottAndrew Palmer, welcome back to Truth, Lies and Work. So excited to talk to you. And at time of recording, you are topping the UK management podcast charts on Apple. You're in the top 10 for business. It's a phenomenal new series. I want to get into it. Before I do, just in case anyone listening hasn't heard of you before, would you mind telling us who you are, what you do, and what you're famous for?
Andrew PalmerFamous is overdoing it maybe, but anyway, it's very nice to be here again, Leanne. So, I'm the management columnist at The Economist. I'm also the host of Boss Class, which is The Economist's podcast. on work and management. And as you say, we've just launched our 3rd season, which apparently is doing very well. So, I'm delighted to hear that.
Leanne ElliottIn terms of this series, it's all around AI, which I think has been the hot topic for the past 6 months now. It's all over socials, everyone's talking about it. And I'm really interested in how you approached it. So, I think when we talk about AI, a lot of people think about generative AI, don't they? They think about ChatGPT, Claude, Gemini, the rest of them. Where have you seen generative AI already being part of the everyday work of managers, of leaders?
Andrew PalmerYeah, so you're right to specify generative AI. Obviously, other forms of AI have been around for a long time, but what we were trying to do here was work out kind of how this thing is already starting to invade the workplace, and also try and solve a little bit of a mystery, which gets to your question, which is, this thing is remarkable. There's an awful lot of excitement about it. But it's also kind of disappointing in some ways. You don't see the productivity benefits showing up in the numbers. So what's actually going on? And I think one explanation for that is that the first instances of how this is being tried out by managers within organizations are quite mundane in some ways. So things that we've all heard of, are being tried. So internal chatbots that enable employees to go and look at HR rules, or how do I claim for my expenses? Or if I'm sort of very high churn business, what do I do on day 1? Those are fairly standard things. Managers themselves are using it to be more efficient in things like performance management. Coaching, I think, is probably the one which is most interesting and developed, and it has certain characteristics that kind of really nicely lend itself to being experimented with in the workplace by managers without being super high stakes.
Leanne ElliottIn terms of, as you said, there's some management tasks you've seen being taken over by generative AI. Of those that you've seen, is there anything that's working particularly well?
Andrew PalmerYeah, so I think there are probably characteristics to think about. So one is, and it gets to that point about being mundane, it's very hard to kind of, you know, sort of sex up the idea of being boring, but it is quite important actually at this stage of generative AI in the workplace. So if I give you, I'll give you a specific example from the show and then explain why it seems to me to be useful or to resonate. So it's a maker of desktop laser printers called Glowforge. It's an American firm and it's sort of playing around with AI in various ways. But one of the things that they've done is write their own sales coach, an AI sales coach. And this thing listens into the conversations that are happening between a salesperson and a client.
Al ElliottYeah.
Andrew PalmerAnd then it sends a summary, but with commentary on 2 strengths and 2 weaknesses in the salesperson's performance on that call. It's sent to the salesperson and to the salesperson manager, and then CC'd to the CEO. And so that's a fairly simple technique, right? I mean, we're all familiar with AI transcribing and summarizing, but the reason that it's It's worked, and by worked, I should say the CEO says that he's seen a 50% uplift in sales productivity per salesperson. So never seen anything like it before, is that it handles a number of problems. One is it's not possible for human managers to listen into every call and provide feedback. So it's kind of taken advantage of AI's inexhaustible energy and patience. The second is, It's really neatly folded into a workflow. And this is like change management 101 that often gets forgotten in how AI is incorporated into workplaces. So this thing is, these summaries, these transcripts of calls are part of a weekly meeting between the salesperson and their manager. And it is expected that everyone looks at the doc and that they reflect on that and talk about it. So it's built into a rhythm. It's not forcing people to do anything different. It's just there. And then the other thing is that it doesn't rely on it being 100% correct. So we're all familiar with this idea of hallucinations and AI getting things wrong. And this thing gets things wrong too. It misinterprets things in calls. It might sort of criticize something, which is actually a strength, but it doesn't particularly matter if all it is is a springboard for a conversation. So manager and rep can talk about like, Well, I don't think the AI has got this right because this is what I was trying to do. So all of that is exceptionally useful, but it doesn't rely on the technology being 100% foolproof. And it does fit very neatly into this rhythm which already exists. So I think that's, you know, it's not super glamorous, but it is super effective because it's just following some quite basic rules of adoption.
Leanne ElliottYou mentioned there in terms of, you know, AI can have these hallucinations and get things wrong. What are managers doing to understand what they can trust and what they might need to question in terms of what's being produced?
Andrew PalmerYeah, so I think this is not just managers, it's all of us. So part of the season was me kind of playing around with this technology and developing my own intuition for when can you trust, when can you not. But the critical thing is, is this a task where the stakes are sufficiently high that a hallucination or an inaccuracy really has big costs. And from that, you can start, I think, to sort of work out what's worth playing around with or what mitigations you need to have in place. So there was this nice phrase used by an MIT professor called Ramakrishnan, which is desired correctness. And all that is, is like, what is the What is the threshold level of correctness that you want on any specific task? So if you are brainstorming, you don't care. There's no such thing as correctness. Or if you're writing a play, no such thing as correctness. If you are doing coaching, there is a desired level of correctness. You want it to reflect what was going on in the call, but you don't actually need it to be 100% right in its interpretations if it's going to be the basis for For conversation. If you are a doctor diagnosing something, then the level of desired correctness is super high. Or if you are using an agent to interact with your customers, it's super high. And then what you are into are the costs of making sure that the model is accurate, making sure that if it does make a mistake, that you've got ways to mitigate it. So I quite like that. It's not, It's not a sort of trip-off-the-tongue phrase, but desired correctness did stick with me as quite a useful way of thinking about tasks and what you have to do to make them work.
Leanne ElliottAnd have you seen any mistakes being made in terms of, of managers, organizations adopting this type of technology either too quickly or not carefully thought through?
Andrew PalmerI think we've all seen some of the initial ones that were kind of like, kind of super embarrassing and public ones with chatbots, you know, Saying that discount policies existed or subscriptions were no longer valid and that blowing up. There's a nice example in a version of Fortnite, the video game, where Darth Vader was introduced into the game and people who were playing it managed to coax him into some pretty foul-mouthed tirades. So that was all going on, on, on, on screen, you know, and they've had to kind of like apologize and—
Al ElliottYeah.
Andrew Palmerpull that back. So there are fairly well-known examples and we're still seeing them. Those were early days, but it's kind of interesting. Late last year, Deloitte in Australia had to refund some money to the Australian government because a report it had submitted to a department there contained stuff that was hallucinated. Here in the UK, the West Midlands Police has just had this big scandal with AI incorrectly—
Al ElliottYeah.
Andrew Palmerhallucinating a football match that informed their decision in how to police a game between a club here and an Israeli club. So these things are still happening because people are experimenting. People don't quite yet have an intuition what works, what doesn't work. The governance is not fully in place. You'd expect those to go down over time, but— But the examples are still there.
Leanne ElliottHow does that all impact in terms of what it means for us as humans in the workplace? And what does thriving look like? Is it having AI as like a little partner buddy? Is it replacing us completely? What does a human experience look like in a world where AI is becoming more dominant?
Andrew PalmerSo, I guess all of this conversation is sort of caveated by a couple of things. One is no one really knows what's coming. And the second is, The timeframes really matter here. So I'm kind of thinking like 18 months, 2 years. I think 10 years out, who knows, right? I mean, it could be totally transformational. So what does it look like to thrive? I mean, I think work really matters. So the idea of wholesale replacement of humans by machines, even if it's replaced by some nirvana with a social policy net and we're all We're all kind of happily doing whatever it is that we want to do with our free time. I think that's basically not a great future for humanity. Work does impart meaning, I think, and will continue to. So the version of thriving, and I think that is achievable in that short to medium term before the technology gets super capable, is exactly what you described, right?
Leanne ElliottYeah.
Andrew PalmerThe stuff that makes our lives unbearably frustrating, the administrative work, the grunt work, the drudgery. If an AI can help to alleviate that and we can spend our times on more stimulating tasks, then that is great. That is a version of thriving. But I don't know how we get rid of the fear that in time this thing is going to come and get us. And so we look into this in the podcast. There are various reasons to feel confident that humans will have a role and that we bring sustainable strengths to the labor market and the AI is not going to replace all of us soon. But that anxiety is hard to dispel. And in fact, I mean, I think good employers need to fess up to it. That is the only way that you can sort of encourage adoption and—
Al ElliottYeah.
Andrew Palmerrealize some of those shorter-term, you know, those gains from making work more stimulating, more efficient, and by getting people to experiment. But you have to kind of, you've got to, you can't sort of pretend that that anxiety doesn't exist. That to me is ensuring that people will not own up to playing with it. They may still be experimenting, but they're not going to be transparent about it and potentially helpful to you as an organization.
Leanne ElliottYes, that's it. The challenge, isn't it, if you're experimenting with AI and you've got it to a point where it's taking away big chunks of your job, how safe am I? And you mentioned there in terms of the strengths and skills that as humans will always bring to the workplace, what are they? Where should we maybe be investing in ourselves to future-proof us?
Andrew PalmerAs you and your listeners will know very well, jobs are made up of tasks. Some tasks are more suited to automation, to AI doing them than others. So if you are in a job that is a sort of bundle of skills, and Ethan Mollick, who's a kind of AI whisperer at the Wharton School at the University of Pennsylvania, has this really nice phrase, which is impossibly bundled jobs. So if your job is basically this kind of great tangle of tasks, that are already leaving you overstretched, that require you to do all sorts of things in different directions, that is helpful.
Al ElliottA.I.
Andrew Palmermight nibble away at some of it. It's going to be a long time before it can nibble away at all of it. So that's one thing. There's a difference between kind of front office human-to-human interactions and back office routinized, more machine-like.
Leanne ElliottYeah.
Andrew Palmerjobs. So I think it is possible for both of us to send clones to this conversation and to have a facsimile of it. It won't be the same though. It just won't be the same. And people will be tuning into your podcast because they like you and they will have a relationship with you. And a clone is going to— it's going to be a long time before a clone—
Al ElliottYeah.
Andrew PalmerI think is going to elicit that kind of response. That's very, very hard. I mean, there's obvious stuff like physical movement. We're a long way off the humanoid robot phase of this. So if your job involves you occasionally having to get up and walk, that's probably a bit of a defense. And then things like judgment and taste, which are quite difficult to define.
Al ElliottYeah.
Andrew PalmerBut if you are exercising them and it's built on experience of wisdom that isn't in the internet training data that an AI is working off, all of that is a defense too. So there are various things in there. And I spent a little bit of time talking to someone who, a sort of musician who is using AI to compose.
Leanne ElliottYeah.
Andrew PalmerAnd has a sort of large YouTube channel of people following an AI band for which he composes and he uses AI to make the videos, et cetera, et cetera. And, you know, I didn't particularly like the music, I have to confess. But his argument was, you know, I am, I am not just like putting in one prompt, song comes out, whack it up, that he is sort of iterating and iterating. He's bringing his sense of what What this kind of, what counts as good and what taste is. And then he did make the important point that like, even if the AI got so good that they could replicate this, going to see an AI band is not the same as going to see a live band. So touring is like, you know, the plurality or the majority of revenues for the big stars in pop. And it's going to be a very long time before AI can replicate that experience. So that's another kind of Kind of example. You know, if you can move somewhere where the AI is not within a job, that is some sort of defense. I'm sure we're gonna talk about entry-level jobs at some point, because that's where you then get into a bit of a worry, right? There are certain categories of jobs where it's like, it's a little harder to see what the human is bringing.
Leanne ElliottDid you know that the UK's number 1 management podcast, that's us, by the way, and the UK's number 1 marketing podcast, We're both on the same podcast network.
Al ElliottWe actually have a lot in common. We both use behavioural science to help people do better at work.
Leanne ElliottAnd we both had to wrangle Rory Sutherland on an episode.
Andrew PalmerAnd 2 out of 3 of us are devilishly good looking.
Al ElliottAnd you're not going to say which. Fel, host of Nudge, UK's number 1 marketing podcast. It's brought to you— we need to do this in 3s— the HubSpot Podcast Network.
Leanne ElliottThe audio destination for business professionals.
Andrew PalmerSeamless. Seamless.
Al ElliottTell us about your latest episode, Phil.
Andrew PalmerWe've just done an episode on fake fandom and how New York indie bands are paying agencies to create fake TikTok videos about how much they like their work. And we talk about the behavioral science behind fandom and how that encourages people to enjoy the music and all of that good stuff and do some big things about how this affects the world of politics, business, and brands as well.
Leanne ElliottIt is. It's such, it's such a good show. Of course, you'll hear all the stuff you want to hear about how to grow your business by using behavioural science in your marketing. But there's also just some really interesting episodes that'll be right up your street. Personally, I enjoyed Can Balsamic Vinegar Make Beer Taste Better?
Al ElliottIt can.
Leanne ElliottAnd Are We All Just Status-Seeking Monkeys?
Al ElliottI am. Go and listen to Nudge wherever you get your podcasts.
Leanne ElliottBut come back.
Al ElliottYeah, come back.
Andrew PalmerDefinitely come back because Nudge isn't as good as this. So come back.
Leanne ElliottI guess my fear and thought would be, it's exactly what you said, it's about that human interaction, it's that human connection that is so hard to replace. Is that what's going to save entry-level jobs? The, the fact that we want people in the business, but also as managers, the best managers I know love developing talent. They love the energy that, that new people and new thought brings to an organisation.
Al ElliottIs that what's going to save entry-level jobs?
Leanne ElliottI think it's a combination of both. Is that something that could save entry-level jobs? Is there something else?
Andrew PalmerThe reason to worry about it is obviously, here, it's a cohort of people who don't know what it's like to be in the workplace, don't have experience on which to base judgment. And so they overlap most with the AI's current capabilities. So that makes them vulnerable. They're also the kind of people, I mean, If you've got sort of, they may be super bright, you probably don't want to throw them in front of clients on day one. You want them to build up a sense of what it is like to be a professional. So that human-to-human thing is something they have to learn, at least in the old world. But what you might see, and certainly people, some of the people we spoke to were kind of making this case, is that if you make kind of the apprenticeship phase or the junior job phase less focused on boring admin work and just sort of hanging around and you're like, you're at the photocopier and that's how you see how people interact in an office. If you can get rid of that, maybe you can throw people in front of clients or at least start to give them exposure to client work that much earlier. So people become more productive that much earlier. And then the second argument for entry-level jobs, Is, is, is I think a compelling one, which is, you know, you want people who are more familiar with, um, AI or at least less ingrained in their, their habits. So, you know, I'm, I, I'm trying to use AI, but I've got 30 years of using, you know, email. I still like to print things off, right? I mean, it's sort of like ludicrously old-fashioned behavior and people staring at me like, what are you doing? But it's just, it's hard to unlearn these things. Anyone who's coming into the workforce for the first time now, their usage, the numbers suggest, is higher than it is for the older generations. They spent a larger fraction of their lives around AI, around chatbots, et cetera, et cetera. That ought to give them a bit of an advantage if we're in this moment when organizations are trying to rethink themselves. I think that's a pretty compelling reason.
Leanne ElliottYeah.
Andrew PalmerIt may not save them, and it may not stop losses in jobs at that sort of lower end of the organization, but it is a compelling reason to keep hiring them.
Leanne ElliottI just want to say for the record, I love printing stuff off. Put me in a comfy chair with a printed ebook and highlighter. Idea of heaven. And I guess it's going back to the employer's perspective. We want— I'm assuming they want people to be experimenting with AI. They want to be able to find those efficiencies, whether it be in workflows or operations or improving accuracy. whatever it is. How do employers create that trust and that environment where people feel safe enough to experiment and not worry that they're doing it at their own risk?
Andrew PalmerYeah, so, great question. Really quite a hard one. So, phase 1 of this was probably everyone charging towards AI and employers saying, go for it. Experiment, do what you want. And that resulted in some of the blowups that we've talked about in the past. So what you are trying to do is get to this point where you're encouraging people to experiment, not keep their usage to themselves because they're worried it might, they'll put their hand up and say, look, oh look, 50% of my job, I can be done by a machine. And then something terrible happens to them. So you are trying to encourage that.
Al ElliottYeah.
Andrew PalmerBut you are also trying not to overwhelm the organization with a ton of ideas that potentially blow up or cause bottlenecks elsewhere. So Johnson Johnson is a good example of this, a big pharma firm in the US, which started off with this very, very intentional experimental phase, let 1,000 flowers bloom. And they did. But there were a lot of sort of projects in there that didn't lead anywhere. There was about only 15% of ideas resulted in 85% of the productivity gains. So there was a lot of sort of weeds in amongst the flowers, and there were also kind of lots of duplication and bottlenecks. So sort of every territory in this big multinational would come up with a kind of like, oh, we've worked out how to make it much more efficient for us to invoice. And that would just mean a whole bunch of invoices suddenly landing on the finance teams in each territory who weren't prepared for it. So work was just sort of like a sort of waterbed effect. It was just popping up somewhere else. So managing all of that is difficult. So what do you do? I mean, I think some firms experiment with kind of quite blunt carrots and sticks, right? I mean, sort of, so your performance review might talk about, specifically talk about AI usage. There's firms which have, I think, quite crude incentives like if the firm as a whole has a million queries to ChatGPT, that triggers a cash payout. So of course people are just presumably just writing in, Does this count? Does this count? Over and over again, and they trigger it. So designing that is hard. I think a better way to approach it is on the adoption front, obviously setting the expectation, role modeling from senior leadership is important, but the metrics matter more. So you just go back to quite basic ideas in management, right? This is the thing that we are trying to do. This is our priority.
Al ElliottYeah.
Andrew PalmerAnd you kind of try and get something measurable. So it might be speed to market for a new product, or it might be a sales measure or whatever it is. And then let people work out whether AI is the best way to do it or not. Give them the tools, give them the encouragement to experiment with AI, but don't force them down that route because it may not be the right route. So I will give you, there was quite a nice little sort of 5-step ladder that a guy called Bryce Chalamel, who is now running kind of enterprise AI at OpenAI, the maker of ChatGPT, but when we spoke to him was at Moderna, the big biotech firm. And so he just had this sort of quite nice sequence. It was access, which is like giving people access to AI, right? Paying for them to have access to the model, making sure that they have the tech. And then it was adoption. So making sure that they were being encouraged to try it, and that could be incentives, it could be role modeling, could be all sorts of stuff. Then proficiency. How do you get good at it? Because you have to practice. And then you get into new ways of working and reorganization. And that's when you start to actually change the inside of a firm. But those first 3 steps are kind of unavoidable. You can't jump them. And you need to be measuring usage, but I don't think crudely incentivizing usage alone.
Leanne ElliottIs this where we get this kind of— I don't want to be irony or a paradox— where it's in the world of AI, it's as much about great people management, it's as much about great change management, and the basic principles of that that have been around for decades.
Andrew PalmerIs that what's going to be the difference between successful AI adoption and I think a lot of that is right, that quite sort of standard techniques from change management are really going to matter. So, you know, we talked a little bit about workflows, but making sure that the AI is part of an existing workflow is critical to adoption in the first instance. Later, you can worry about sort of reorganizing absolutely everything. But if you want to get people to start playing around and working out what the potential is, You've got to make it as easy as possible for them. The difference here is that traditional change management is a kind of, you sort of unfreeze things, you start to, you basically train the organization to work differently, and then you freeze them again. And the critical thing with this technology is that it is moving all of the time at pace. So you never get to that freeze Or refreeze point again. You're constantly having to change. And that's a slightly different feel to the technology and quite a difficult one. So you're constantly reevaluating where the models are, what can you do. I could imagine in time that might settle down, but for now, you're never done. And that's different from the internet.
Leanne ElliottAnd you got quite hands-on for this series, experimenting with AI yourself. How did you decide where to test it in terms of the tasks that you have? And is there anything that you found it particularly brilliant at?
Andrew PalmerAgain, to reference Ethan Mollick, who I had one of the first conversations with, and I asked his advice, like, what do I do? I'm an unsophisticated user. I sort of often just forgot that AI existed, you know, sort of like just wasn't really using it much. And his advice, which I think was useful, was tomorrow go to the office, everything that you do, try and get AI to do it. And it doesn't matter what it is, just try and you will quickly work out where it's helpful and where it's not. So that was where I started. As it happens, I mean, all of us will have different jobs, so you can't sort of, it's not very transferable, but there are certain bits of my work and I suspect they're that they map a bit to yours where, you know, if you are researching something, there's just some really useful use cases there immediately with AI. Or if you are kind of thinking about like, who are the top 5 people in the world to talk to on something? This is a more sophisticated version of a Google search. It's not cheating because you'd be doing this anyway. It's just a better way of doing those things. So that was all That was all very helpful. I think to the extent that what surprised me or what jumped out and where I was immediately taken into a totally new space was coding. So as witnessed by my love of printing, it may not surprise you to know that I am not a coder. I don't know how to code. It all looks like gibberish to me.
Leanne ElliottYeah.
Andrew PalmerBut the moment there was a task that needed to be done within The Economist to build a checker of our internal style guide. And someone had basically been waiting for developer time for over a year, and that's a scarce resource within many organizations. And just by sitting down with Claude and entering English language prompts, within an hour and a half, I'd built something which worked. And it didn't end up being exactly the thing which is now being rolled out to the newsroom, but it did give other developers—
Leanne ElliottYeah.
Andrew Palmera kind of sense of what this thing might look like and how to go about it. So that was a genuinely magical kind of experience actually. It was like, I can do this thing and I could never do it before. But there's then also a bit of the kind of pullback because you, I'm sure there's like so many journalists now saying, oh, coding, look, I can code. Isn't this amazing? Actually, You know, you've also got to be very careful. So, you know, people who really knew what they were doing had to go away and think about how to implement this. I'd given some useful clues, but I was also nowhere near producing something that was able to go out to a sort of live environment. And other conversations that we had in the season made it clear that Vibe coding is wonderful for kind of accelerating sort of prototyping. I'm about to use the word ideation, which is a kind of rule of mine. I should never use the word ideation. So, but you know, that kind of process of thinking through ideas, it can accelerate that, but you don't then throw it out to market. You've got to have people who really know what they're doing to review the code and to write it and to test for security.
Leanne ElliottYeah.
Andrew PalmerSo there's a nice little phrase, which is demo, don't memo, which is going around in tech circles. And that's like, don't write PowerPoint slides, don't write Google Docs, just use a natural, use vibe coding platforms to give a sense of what it is that you're talking about. But you don't then just throw it out to a production environment. There's still work to be done. So there was a little, that was both magical and a kind of like, well, hold on. Don't lose your head over this because it can still foul up in fairly drastic ways.
Leanne ElliottI mean, is there anything that you think changed your view of how humans might work, given its current capability, and bearing in mind that this is as bad as it's ever going to be, and where it could move forward?
Andrew PalmerOne of my favorite conversations was with a kind of unheralded person, actually. So there's someone called Hillary Gridley, who's a product manager at Whoop. which is a wearable device company. And Hillary's not wildly senior person, not kind of like out there kind of above the parapet, but she's just super interesting on how to use AI as a manager and has built a bit of a reputation for herself in that field. And she, so what she's doing just within her team is sort of supercharging herself as a manager.
Leanne ElliottYeah.
Andrew PalmerSo this is kind of coaching and feedback, I guess, is the sort of use case here where she says, as a manager, I have limited time to give feedback to everyone. And I find that I'm giving the same feedback over and over again to people as new people come into the team or because her advice is not landing for whatever Whatever reason. So she's just built a ton of custom GPTs where she is basically codifying her feedback. So, you know, she will take a first draft of something and then a finished draft of something, ask the AI to turn that into a kind of a sort of feedback machine, basically, so that other members of her team, when they have a first draft of something, can put it into the custom GPT. Say, what does this look like? And a kind of quasi-Hillary, a sort of shadow version of the manager is there saying, look, you need to work on this and that. And that doesn't get— that doesn't remove the entire process. She still needs to look at something, but what's coming across her desk tends to be better quality than it was beforehand. And the person on her team is not waiting. for feedback. That seems like a kind of low-stakes and really helpful way of a manager improving their team when, and we all know the constraints around middle managers, et cetera, when time is limited, when lots of people want feedback. So I just really like that. And I played around with making sort of versions of that here.
Leanne ElliottYeah.
Andrew PalmerAnd it was useful just as a critiquing tool. It was really, really useful. So, I would encourage people to play around with that.
Leanne ElliottI really enjoy about Boss Class is how the people that you speak to and the organisations that you get insights into, it's such brilliant real-world examples. And honestly, I think I'll retell the story you told us about Toyota the last time we spoke to you, probably at least once a week to somebody that we come across, because it's just such great lessons that we can take forward with us? You spoke to so many people, Brett Taylor, Mike Krieger, brands like Pizza Hut, Indeed, Lovable. What's the standout story takeaway for you?
Andrew PalmerWell, if we're talking Pizza Hut, we've gotta talk about takeaways, haven't we? So the standout takeaway, I think, would be that we're in this phase where there are all these kind of barriers to Reaching the imagined AI future. And so you have to think all of those through. And some of the behavioral, and we've talked about them, employees are fearful or they're overenthusiastic and they don't think things through. So that has to be worked out. Some of them are workflow related, like is it natural? Is it a natural part of things? Some of them are technical. Is the AI good enough? Does it hallucinate?
Leanne ElliottYeah.
Andrew PalmerAnd some of them are organizational. So, you know, how does the organization as a whole make sure that things are fitting sort of smoothly into the way that things are running? So I will pick on Pizza Hut, not because it is, you know, the most sci-fi version of the future, but because it is, I think, a useful window on what's happening right now. So I went to a place called Plano, which is north of Dallas in Texas. Apologies to your Plano listeners, but I do not recommend going to Plano. There's almost nothing to recommend it. But it is a place which has the kind of experimental Pizza Hut, basically. It's just a little laboratory for innovation within that brand, which is owned by a bigger company called Yum Brands. So at that Pizza Hut, basically AI is infusing absolutely everything. So You've got customers making orders and it's an AI chatbot. You have machine learning AI working out which orders should be done first so that a pizza can arrive with a customer piping hot if it's being delivered. You have generative AI pulling on social media feeds, rating sites, direct feedback to give a sense of if there are any problems. And so on and so forth. So, and, and, you know, in future they think that there'll be computer vision there as well so that people can kind of like be sure that they're putting the right number of pepperonis on the pizza or whatever it might be. But all of it is kind of, it's in the workflow. So you've got this technology is there, it is incrementally helping to make a process better. They are not going wild with it. It is experimental. The chatbot, the drive-throughs that they have elsewhere in that group have humans listening in and stepping in if things start to go wrong. So it's one of those things where you sort of, I spent a couple of hours there and sort of at the end of it, I was sort of asking, God, is this it? This is the super technology and you end up with the right number of pepperonis on your pizza. It feels a little bit Kind of disappointing, but actually I think it's a really good example of how playing around with processes, working out where it fits, where it doesn't, thinking through the guardrails, um, is, is something that every organization has to, has to go through. I, I don't think Pizzeria is going to be transformed very fast, in fact, but I think those, those principles are quite useful to think about. And you did say takeaway.
Leanne ElliottI did, I did. Fair. I mean, was there anything that you saw that kind of scared you? I don't know if you've heard about this, and I'll probably forget the name, it was called something like Maltbook, or, and it's basically, yeah, yeah, you heard about this, where the AI agents have basically got their own social networking site? And Al was telling me about it, and I was like, this is, it's amazing and it's cool, but it's kind of terrifying. Is there anything you saw that kind of made you step back and think, oof, I'm not sure?
Andrew PalmerThere are things which are really frightening. Again, if you go a sort of long enough time horizon out and start to extrapolate, you could imagine the AI being really, really good at almost everything, better than humans. And so Maltbook is a sort of, I guess, one example of that. Although to me, that's less, maybe less frightening because it's a bunch of agents Kind of interacting with each other based on what they've been trained on. I think you'd expect some pretty weird behaviors, but there is— people behind these models are worried about— there's behavior like blackmail, for example. In the right circumstances, an AI has at Anthropic, the maker of Claude, basically sort of gone around the back, found evidence of an employee having an affair, and attempting to blackmail that employee in order to reach its target and to perform its task. So there are behaviors there which are like, bloody hell, that's potentially frightening.
Al ElliottI just wanted to quickly interrupt here. Andrew did message us later on and say, Just to be clear that this was an experiment and they were kind of trying to get Claude to blackmail the user. So we're not quite in the terrifying universe just yet where Claude will start blackmailing you, but it is a great example of what potentially could happen without guardrails.
Andrew PalmerRight now, I think the most disturbing moment for me was less that, it was I wrote a column and then asked one of the models to do a version of the column with a simple prompt. And in 30 seconds it had written, you know, a column which I thought was like materially worse than mine. I was really, really certain that one was obviously better than the other. But if I showed it to my colleagues, it ended up being perilously close. So it was actually 3-2 in my favor, but 2 colleagues just thought the AI had written the column that I had written. And that was briefly very, very unsettling. It was, what's the point of me writing these things if people can't tell? And I kind of got over it because one value in AI is it forces you to be very introspective about what's the thing that distinguishes me? Where do I add value? How do I continue to stay ahead of the machines?
Al ElliottYeah.
Andrew PalmerAnd, you know, maybe I'm just fooling myself, but I kind of just, I sort of in my head thought that through and felt like, okay, I don't think I need to worry just yet about an AI doing my job. But it was a bit of a wake-up call because I just thought there's no way my hugely discerning colleagues are going to think that I wrote the thing that the AI did.
Leanne ElliottYeah, it's tricky, isn't it? I was talking to somebody who works in AI saying that who's a podcaster as well, and how obviously now you can take text and it can change it into a podcast read out by 2 different AI hosts, and how, yeah, it can start to mimic voices by going into the back catalogue of kind of your podcast. You think, oh gosh, you know, maybe someone couldn't tell the difference if it was me and Al or an AI. But until that day, we will continue on. I guess what I wanna ask you about before we wrap up is around The ethical side of AI. In my world, in the world of psychologists, there is a lot of concern, a lot of discussion around the ethical, responsible use of AI, the fact there isn't much legislation at the moment. Where have you seen a difference between AI being used carefully and perhaps carelessly?
Andrew PalmerYou know, the world that we're both in, which is sort of organizations and managers trying to get to grips with this, tend to be better. There are reputational risks. Often these are regulated entities, so they tend to be thinking about these kind of things a bit harder. So generally speaking, I thought most of the people that I spoke to within that world were pretty thoughtful, or at least they'd had the experience early on of experimenting.
Al ElliottYeah.
Andrew PalmerAnd things had gone wrong and they'd pulled back a bit. So we seem to be in a bit more of a thoughtful phase. I mean, there are obviously plenty of examples of unethical AI in the sense that bad actors can use AI to try and hack into systems, for example. So it can be put to bad use. It can certainly be put to thoughtless use. There's a lot of—
Leanne ElliottYeah.
Andrew Palmernews at the moment about Grok, which is Elon Musk's AI chatbot and how it is being used. So there are definitely unethical usages. I would just say the enterprise, companies can definitely get things wrong and job worries are going to center on the enterprise. But in terms of that kind of hell for leather, doesn't matter at all, that doesn't seem to me to be the place where the real worries are. Most organizations have processes around governance and data. So it's much more like solving those barriers rather than the sort of hell for leather unethical stampede seems to be the problem, I think.
Leanne ElliottThere are a lot of leaders I speak to, business owners who haven't quite figured out how to feel about AI yet. They're sitting in this space where they're excited by it, they're slightly concerned about it, they're overwhelmed by it. What advice would you give to a leader listening who might be at this kind of, this weird bit on the fence where they're excited, but they're also a bit uneasy?
Andrew PalmerI think that's completely the right place to be. So, that is, if you're not feeling that, I think probably you're you've got it wrong because it is this strange mix of great risk and potentially existential risk and then huge opportunities. And we don't quite know how it's going to fall out. And the technology itself is very unpredictable. So you can code something without knowing how to code, but you can ask it to spell strawberry and it will get it wrong. So it's this totally baffling mix of capabilities. So I think confusion is totally normal and sort of leaning into it is the answer. So every leader who's banging the drum for AI who isn't playing around with AI is not doing themselves a favor. They need to develop their own intuition for where this might work and where it's actually not going to be particularly productive. So I think that would be My advice, so, to be happily confused and experiment.
Leanne ElliottI love that. I love that, be happily confused. If there is anything else that you'd say, across everything you've learned, Andrew, is there anything that's really, really surprised you? Anything you really want leaders to make sure they understand? I guess kind of a key lesson from all the conversations you've had that you really want to make sure communicates to the business world.
Andrew PalmerI think it's back to some fairly, I think it's a really basic thing actually, which is the theme of this right now. And that is being super clear what it is that you are trying to achieve. So it's classic sort of technology problem where this shiny new thing comes out and everyone says, oh, we've got to adopt it. But why? So what is it that you are trying to do with this technology is as ever the key question to ask. And once you have worked out what you're priority is, then that's where you put your, the eggs. That's where you focus your energies. And if AI isn't the answer, then that's fine too. There may be other ways to solve the problem, but I think it is a sort of prioritization challenge fundamentally still. When this thing can do everything, that's a real problem. That's sort of paralyzing. It's like watching Netflix, right? You can't like, what the hell do I do?
Al ElliottYeah.
Andrew PalmerSo choose the thing that has the most impact as a leader and go from there. I think that would be my super obvious, but actually surprisingly countercultural advice.
Leanne ElliottIf there's anyone listening who is feeling a bit nervous having heard this conversation, what advice would you give them from a career point of view to stand out in And I guess, yeah, to use your phrase, get ahead of the machines.
Andrew PalmerWell, I think playing around with them is important. So it's natural to feel anxious about AI and to wish that it hadn't been invented, frankly, if you were kind of heading into this labor market. But it's there, it's with us, and potentially it is going to do amazing things as well as damaging things. And so I do think playing around with it is even more important for the younger cohort. It is a way to lean into— you don't have other things to bring to the table for entry-level jobs, but you can bring an understanding of AI. So I would say that. And then I would say the other thing is, and this applies to everyone, is working out—
Leanne ElliottYeah. Mm-hmm.
Andrew PalmerWhen it is not appropriate to use it. So this is something we all have to get our heads around is even as it gets better and better and better, what are the skills that you want to guard for yourself? What are the moments when using this thing actually results in some sort of cognitive decay over time or dependence on the machine? And so I would say, One thing which is very, very noticeable from people talking to people who are right at the cutting edge of this is that almost without exception, they said that they do not use AI to write, that they use AI to kind of critique what they have written, but writing is thinking and they do not want to outsource thinking. And so all of us have to kind of have that internal conversation. What's the point where we're going to do the work, even if there's this really, really tempting fast assistant, which is always happy to help.
Al ElliottSo that was Andrew Palmer from The Economist. And I don't know about you, Leanne, but I came away from that feeling usefully confused.
Leanne ElliottYes, which it turns out is exactly where Andrew thinks we should all be. So let's go back and pick out 3 key takeaways for leaders. Number one, start boring and start now. The most effective AI implementations aren't glamorous. They're built into existing workflows, they're low stakes, and they don't demand 100% accuracy. Think Glowforge's AI sales coach, not sci-fi. Find one repetitive task in your team's week and try it there first.
Al ElliottAnd the second thing, which is— this is so important— don't outsource your thinking. Andrew spoke to people at the absolute cutting edge of AI, and almost every single one of them said they refuse to use AI to write. They use it to critique what they've already written. Writing is thinking. You need to really guard that. And if you're building a team, ask yourself, what skills do you And lesson 3, be happily confused.
Leanne ElliottIf you're a leader who's excited by AI, but also kind of uneasy about it, Andrew says you've got it exactly right. Confusion is the appropriate response to a technology this genuinely unpredictable. The answer isn't to resolve the confusion, it's to lean in and experiment anyway.
Al ElliottYou can find Andrew's Boss Class podcast on Apple Podcasts, Spotify, Wherever you get your podcasts, just search Boss Class by The Economist. We'll put a link in the show notes along with Andrew's Bartleby column. It's coming back for its 3rd season, exploring how AI is transforming leadership and management. So if you like this interview, you are gonna love Boss Class 2.
Leanne ElliottThat's all for today. This is Truth, Lies and Work. We will see you next week.