Boss Class 1. Fat layer of humans

Why get AAA? Bad luck. Stuff happens. Triple A's got your No matter whose car you're in. Join today at triple A dot com. Triple A. Expect something more. Roadside assistance is provided by independent facilities contracted by Triple A. Coverage in taxis, limousines, and other writing. 2024 AAA U.S. Market Track National Surveys. Offers, terms, conditions, and policies are subject to change without notice. Visit triple A.com for details.

Have you been waiting for the perfect time to upgrade your tech? Good news, the wait is over. Dell Tech Day's annual sales event is here, and we're celebrating our best customers with fantastic deals on the latest PCs, like the Dell 14 Plus with Intel Core Ultra Processors.

We've also got incredible perks like Dell Rewards, fast free shipping, premium support, price match guarantee, and more. And while you're upgrading your PC, you may as well go all out, because we're also offering huge deals on our premium suite of monitors and accessories. You know what that means? That's right. You can get a whole new setup with amazing savings. Clearly, this is a sale you don't want to miss. Visit Dell.com slash deals. That's Dell.com slash deals. The Economist.

Hi, it's Jason Palmer here. The real one, not some AI simulation. You never know these days. I'm here to introduce season 3 of Boss Class. You clamored for more, now it's here. Over the next three weeks, we'll be publishing two episodes each week in the same place you find the intelligence. Meanwhile, the weekend intelligence will be out of office while we publish boss class. They'll be traveling and reporting and gathering the compelling stories you'll hear when it's back.

Season 3 of Boss Class is all about generative AI at work, something that most of us find at least a little terrifying, or should. Andrew Palmer, no. is back as our guide. He writes Bartleby, our column on work and management. He knows he should probably be using more AI at work, but he isn't sure where to start. BossClass asks why AI is so frighteningly good at some things, but equally frighteningly bad at others, and what that means for the future of all of our jobs.

Now, like purveyors of other addictive substances, we're making the first dose free. But for the rest of the series, you'll need to be a subscriber. Just search for Economist Podcasts Plus to find our very tastiest offer wherever you are. Uh hi, who is this? Who am I talking to? Hello, I'm Andrew Palmer, a senior writer at The Economist and the host of the Boss Class podcast. Hello Andrew, this is totally weird.

This is totally weird. I'm Andrew Palmer, senior writer at The Economist and host of the Boss Class podcast. That is my digital clone. It's been trained on my writing and my voice to answer questions about management and the world of work. According to my Bartleby column, the workplace, day by day, is a theatre of mild agitation. I wrote that suspense can come from the smallest things, like entering and exiting meetings or presenting when the clicker doesn't work.

Yes. It only has a knowledge base of what you've written or special. Ruth Berry created this monster. She's a colleague here at The Economist and is part of an internal team experimenting with artificial intelligence. How long has it taken you to do this? Uh this was an afternoon. Right. Okay. Very fast. Yes. Okay. To make the clone, Ruth uploaded dozens of my columns into ChatGPT, and then hooked it up to an AI-powered voice generator, which had been trained on recordings of me.

Thanks to this podcast, there are quite a few of those. This is what makes you a very good subject. Subject. Subject. Object. Labra. Having fed it samples of my writing, she also asked the AI to come up with a prompt describing my general vibe. So warm, dry, British wit, lightly ironic, never gag heavy, apparently is what is put down there. Never gag heavy. Normally this would be a good point for a gag, but now I'm not so sure.

Time to ask my doppelganger a loaded question. Can you tell me what most irritates me about office life? Hello, I'm Andrew Palmer. According to a recent article in The Economist, one of the most common irritations is when colleagues take credit for other people's ideas. Turns out it's also pretty annoying when your replica does it. Although in fairness, it's hard to stay angry with this particular robot for very long. Have you got any advice about how I could negotiate myself a raise?

I'm Andrew Palmer. When negotiating a raise with your boss, Jim Sabinius from Harvard Business School suggests doing your homework to find salary ranges. My voice clone is a party trick. It took Ruth an afternoon to put together. It only has one topic of conversation. It introduces itself a lot. But think about that for a moment. After just a few hours' work, I am talking to a version of myself. It's definitely not me, but with more work and better technology, it might be much harder to tell.

Because if AI was just a party trick, then nothing about this moment would make sense. The trillion-dollar highs of the biggest tech firms, the fear of a jobs apocalypse as the machines become ever more capable. The efforts that firms everywhere, the economist included, are putting into understanding AI. That is the strange backdrop to this third season of my podcast on work and management. Plenty of people believe AI is going to transform the future of white collar war.

that machines will outperform people in job after job. But back in the present, that transformation can feel a very long way off. Do you have many concerns about the voice cloning and the use of AI? Uh so I I think it is so obviously like not if I if I started every single response to a person with I'm Andrew Palmer, then I would be more worried. I'm Andrew Palmer.

For the past few months, I've been in search of people who are at the frontier of generative AI in the workplace. I've spoken to entrepreneurs, academics, engineers and managers from around the world. People who are testing its strengths and discovering its weaknesses. And with their help, I've been trying to answer two big questions. What's the best way for bosses and employees to use the technology at work right now?

And what should we be doing to prepare for the future that's barreling towards us? Over the course of the next six episodes, I'll be sharing what I've found with a warm, lightly ironic British wit that's never gag-headed. From The Economist, this is Boss Class, Season 3, Episode 1, Fat Layer of Humans. With a long enough time horizon and a good enough AI, I just don't think there are that many situations where you prefer humans.

Tom Blomfield is paid to think about the future of AI, and he thinks that it's coming for almost everything. Medicine is a really good example for me. I just turned forty, I'm become more health conscious.

Every blood test I get I put into AI, every doctor's prescription I put into AI, and I'm like way more informed about my health. It's just I want the right answer. I don't care whether it comes from a human or an AI. I just want to access the best information possible and I want the best advice. Forget bedside manor. Enter bedside scanner. Tom has some pretty solid credentials. He made a fortune co-founding Monzo, an online bank with a clever app that's now used by one in five adults in Britain.

He left Monzo in 2021. These days, he's a partner at Y Combinator, a famed startup accelerator in Silicon Valley. Companies that have come through its doors include Airbnb, Reddit, Stripe and Twitch. Over half of the current intake are AI startups. If you want to know what the true believers think is going to happen with Gen AI technology, this is a good place to come.

So I was primed for some dramatic predictions when I pitched up at YC's new campus of converted factories and warehouses in a trendy waterfront neighborhood in San Francisco. Even so, I was surprised by what he told me. The big change for us in the last two years is that very, very high achieving young people. have got it into their heads that this AI thing is real and therefore they only have a limited period of time in which to accrue wealth.

to position themselves in this you know, the stratus of society they want, which I think is a like dystopian, weird thing to um to think and I don't think is true. But it is it's like this kind of mind virus that's got into university students, especially in the US. And so people are dropping out at much, much, much higher rates.

And thinking they need to start a startup now because all the ideas are gonna be gone, you know, when the super intelligent AI is out in X years. Wow. I hadn't I hadn't heard that before. Okay, so well, there's a sort of five years to kick become a billionaire, basically. Wow. Oh yeah. Yeah. Tom himself doesn't think we'll be living out Brave New World by the turn of the decade, but he does believe that a sea change is already underway.

I come from the perspective of I'm working with these tiny startups who started with zero revenue a year ago and um they are targeting that one percent of problems that could be solved in an enterprise right now and they're going from zero to ten million of revenue. faster than I've ever seen companies grow in my twenty years in this industry.

And their deployments are getting very high engagement and then expanding. And uh from the perspective of a big enterprise, maybe yeah, maybe they only are only solving one or two or three percent of the problems right now. But I just bet on that expanding much, much faster. Last year Tom stirred up some controversy with a prediction.

He tweeted that software engineers are like farmers, while AI is like a combine harvester. The world is going to have a lot more food and a lot fewer farmers in very short order, he wrote. So what did you mean by that? Um basically that AI is going to automate vast swathes of the job that software engineers used to do.

And you'll have a human overseer probably, but I think the requirement for human input will go down very, very dramatically. And I think the overall output of software will go up very, very, very dramatically. And I think that will create a huge Sort of surplus for the world. Writing computer code is one of the things that Gen AI systems can already do very effectively. And that has implications for white collar workers everywhere. You go into any business in the world. And

The level to which they're run on on basically Microsoft Excel is sort of astonishing. You know, you just like huge, huge lists and humans are filtering and sorting and copying and pasting from PDFs into Excel, back into email. I think AI coding tools will enable anyone to take those very, very repetitive operational tasks and just automate them. In the short term, Tom thinks that ought to be great news for anyone working as a desk jockey. AI tools will take drudge work off their hands.

But over time, it'll do more than that. I have very, very high certainty that almost all knowledge-based work will be done at a higher level by AI within our lifetime. Think of everyone you know whose work is done on a computer. That's who Tom's talking about. Instead of humans, agents, autonomous AI workers, will be turned loose on complex tasks. And I think we'll get vast improvements just by the way. Just by layering multiple agents on top of each other.

Yeah. And presumably right at the top somewhere there's some thin layer in this future. There's some there's some thin layer of humans applying a judgment about like this works, this is good, this is not. Yes. Right right now there's a very fat layer of humans.

And I guess my argument is just that layer gets thinner and thinner and thinner. And does it ever go to zero? Like, yeah, I don't see why not. What is the timeframe over which that happens and how thin does that layer of humans get? And what do the rest of us do? Why give it a little bit more? Bad roads, bad tires. Out of gas, locked out, or in a ditch, stuff happens. With twenty four seven Roadside Assistance, Triple A's coming to the No matter whose car you're in. Join today at triple A.com.

Expect something more. Roadside assistance is provided by independent facilities community. Coverage in taxis, limousines, and other rides. With no fees or minimums on checking accounts, it's no wonder the Capital One Bank Guy is so passionate about banking with. If he were here, he wouldn't just tell you about no fees or minimums. He'd also talk about how most Capital One cafes are open 7. to assist with your banking needs.

Even on weekends, it's pretty much all he talks about, in a good way. What's in your wallet? Terms apply. See CapitalOne.com slash bank, Capital One NA member FDIC.

Meeting Tom left me feeling unsettled. According to him, AI is on a trajectory to put us all out of work. But even he admits his view is bound to be conditioned by what he does and where he lives. In the here and now, and away from the sunshine of Silicon Valley, it's hard not to feel much more confused. Because if Gen AI is so powerful, why does it so frequently disappoint? I needed someone to give me a different perspective, and I knew just where to turn.

I mean the first thing that you have to know is kind of the economist we embrace change. I mean that's part of our ideology kind of. W w we think technology is good and should be used. Ludwig Siegler is a seasoned tech editor who's been at The Economist for over 25 years. He's an unusual blend of sensibilities. Part Californian, part German.

In 2023, he was put in charge of a new editorial initiative at the paper, to figure out how to implement generative AI tools across our newsroom. It's the same team that Ruth, who made my voice clone, works on. Their work has been driven by a question that will also be helping to guide this series. So what we needed to do is to figure out what is the skill frontier of these models? How can we use them?

How can we not use them? Uh uh I think experts talk about the kind of the jagged frontier of AI, kind of they're very strong, very good at certain things and other times they surprise you in how weak they are. The Jagged Frontier Keep that phrase in mind, it's a helpful one. Because the border between useful and useless AI is not a straight line, it's a mountain range. For some tasks, the technology has already climbed to high altitudes. For many others, it's still putting on its boots.

This can make the experience of engaging with Gen AI Somewhat trying. I would say with Gen AI, you always teeter at the edge of disappointment. It's not an easy process. Actually, it's the nature of the technology to be imperfect. And uh human workflows or human organizations are difficult. They're set in their ways. Inertia is kind of the default. Ludwig's experience is not unusual. Plenty of firms have failed to see any returns on their Gen AI projects. Disappointment is common.

But so is the fear of being left behind. Some companies are managing to extract value from the technology. News of major AI breakthroughs keeps flooding in, and everyone is racing to learn from their failures. We'll be hearing more from Ludwig later, but if you take one piece of advice from him, let it be this. It's normal to feel confused about AI. Desirable even.

There's days where I think this is really great and it's gonna change and the opportunities are bigger than the threats. There are other days where I think this is just crap and it's never gonna work'cause I'm torn in in this way and I think that's a pretty good thing to have. This is where Ludwig's blend of Silicon Valley optimism and Saxon realism really comes into its own. If I were just kind of the German skeptic here, that wouldn't work. But if I'm kind of

The hipster or hipster that thinks this is the best thing since sliced bread or something. It wouldn't work either. I think it's very healthy to be torn with this type of technology. Healthy, maybe. But also disorientating. After speaking to Tom and Ludwig, I was reassured that my sense of confusion about AI is normal.

But that left me with a lingering question. I'm a white collar professional with no background in technology. And although I use AI once in a while, I also occasionally forget that it exists. I've never properly got to grips with it. So what should I do next?

I could tell you as a writer myself, if you're not using deep research tools in writing, you're making a mistake. And on that front, who better to sit down with than one of the most influential thinkers in the field? A man who helped coin the term the jagged frontier. Hi, I'm Ethan Moloch, a professor at the Warden School of the University of Pennsylvania and author of Cointelligence and someone who studies and thinks about AI. And yes, a man who also speaks very fast.

If you listen to podcasts at higher speeds, you may want to slow things down a tad. Like Ludwig, Ethan also thinks that it's okay to be confused about AI. It's early days. I mean, this technology hit the world in a way that people noticed in twenty twenty three essentially. And we're a couple of years in. So it's always weird to me. It's like, where's the giant productivity gains? Like,

I'd be very surprised if we saw every it's a mess everywhere, right? We're all trying to figure all this stuff out. Ethan's best selling book, Cointelligence, isn't about machines replacing jobs. It's about the disruption happening within jobs and how to capitalize on that disruption. It's about joining the bots.

I think you don't get AI unless you've had three sleepless nights of sort of anxiety, right? So I actually, you know, stayed up a night being like, what does this mean? I mean, this is producing human seeming language. So what does this mean for me? What does it mean for my kids? What does it mean for the careers I teach people to do? And so there's a moment I think everybody has when you work with these systems that they're deeply unnerving in some way. Ethan has clearly got past this phase.

So I wondered, how exactly does he use AI in his own work? And how does he decide what not to use it for? I do a lot of writing. I never have the AI write first drafts of anything I do and in fact don't engage with an AI until after I write a draft, but then engage a lot with the AI as an editor. And I've never thought having an editor

was a negative thing. But on the other hand, for a task I care less about or that's important but less interesting, like if I have to fill out a form, I'll have the AI do the first draft of that because that scatters my attention and I can focus on things I care about. So your kind of red line, your threshold is sort of things that you regard as being sort of more integral to your identity and or things that you are better at.

Or things that help me succeed, right? As somebody who writes a lot, if I don't write drafts out I don't think through a problem. And then I guess one other problem is just the performance of the AI. Can you trust it to do a good job or is actually the the upfront investment in sort of checking the work worth using it? So what I can tell you is hallucinations are dropping over time overall. And the question is what's the threshold at which you stop worrying so much? Or when do you stop checking?

And I don't think we have answers on that yet, but we're still figuring it out. It's good at some stuff, bad at some stuff. And the only way to know what the frontier is is to use in your area of expertise because then you can start to figure out that shape. People may also be wary of embracing AI in their work, in case bosses jump to an unpleasant conclusion.

I worry about management a lot with AI. Um, and I worry about it for a couple of reasons. One of which is what do you do with an efficiency gain over the last 20 or 30 years? You fire people, right? That is the that's the incentive to do that. That is not what you do in the middle of the industrial revolution. What you do is you have a capacity gain, you use the capacity gain to gain an advantage because you have new capacities you didn't have before.

In Ethan's view, companies that see AI as a cost cutting tool are fundamentally misreading the moment. Rather than replacing employees with Gen AI systems, managers should be thinking about how to reinvent their organizations. Which brings us to his second worry a failure to reimagine how work gets done.

To step back, I think this is an R and D problem. And I think organizations are gonna have to spend money on organizational R and D, which may include not cutting people the way they thought they would. Do you have examples of organizations which are starting to think through those sort of innovations in form?

Yes, I have seen a very large organization which has started to experiment in some other divisions with taking their most senior engineers, pulling them out of the IT department, and then having groups of three that do the work, a senior subject matter expert, a senior engineer, and maybe a marketer.

or a strategy person and they literally build new products from scratch in a week that would have taken six months and a teams of twenty to do. So playing with organizational form and moving the boundaries of the organization closer together is something I'm seeing a lot of. The ways in which AI will change the work of managers and the shape of organizations are going to be a big theme of this series. But I still had a burning question for Ethan, one concerning a certain individual.

We obviously don't know each other, but based on this conversation, um and my self confessed status as an unsophisticated user, what would what homework would you set me? So I would first of all ask you, are you using it for everything? Have you tried using it for every part of your job? No, it's completely sort of sporadic and scattershot at the moment. So tomorrow, when you go into work, use it for everything.

you know, uh literally try and have it do every task and then but actually have it try and see how far you get and you'll find that jagged frontier. It's gonna be really good at some stuff, bad some stuff you you don't expect. And if nothing's freaked you out, you're not being ambitious enough.

Alright, um just testing that this is working. Sounds a bit noisy, hold on. I took Ethan's advice. And, begrudgingly, I also took my producer's advice. to record my progress in an audio diary. And what I thought I would do While I'm sitting here in my perch at the Adelphi in the Economist's mothership is a first go at integrating AI into my weekly column tasks.

So it captured an exchange that may come back to haunt me when the superintelligent machines take over. I'm afraid I already hate chat GPTs. Persona. I might I might ask if I can change it. Hold on one sec. Uh right, oh sorry Chat GPT, I think you were listening.

Absolutely, I'm here and ready to roll with whatever you need. If you'd like to tweak how I chat or change up my persona a bit, just let me know. I'm all ears. Oh my god, that was embarrassing. Um with my fate already sealed, I figured I might as well press on. And as the hours went by, I clocked up several small scale wins. Using Chat GPT to do a first pass at a review of academic research, rather than trawling Google Scholar.

brainstorming ideas for potential interviewees with Anthropics chatbot, Claude, and using a Google product called Notebook LM to summarize proofs of business books. So you know I guess the the sort of characterization of AI as a sort of very competent research assistant rings true and that is amazing given the speed and thoroughness of a lot of what is done.

But I also had several interactions which were less impressive. My request is: I am the host of a podcast about management. We are producing a series. Sorry about that. The system didn't pick up what you said. Could you repeat that? Sometimes the models hallucinated, confidently telling me things which turned out to be inaccurate. At others they glitched out, as Claude did during this exchange. Yes. I am the host of Go ahead. I'm listening. What are you producing for your management podcast?

I shouldn't really lose my temper with a nonsentient being. I will try again. Up to this point, it was all more or less as I expected it to be. Helpful in some areas, frustrating in others. And then, just as Ethan predicted, I had my freakout moment. A few days earlier, I'd written my weekly column, a satire, poking fun at a mainstay of office life, brainstorming.

The piece was 700 words long, structured as a kind of mini-stage play. It featured a group of people in a conference room, throwing out names for a new product. Here's a short excerpt, read by synthetic voices from the same service that created the I'm Andrew Palmer clone. Let's get going. We've all had a chance to think of some fresh names for our new value-added membership service. So let's write our favorite ideas on the whiteboards and then we'll review them.

Sound of breathing, sharpies, and paper. Okay, let's take a look and see if any themes emerge. I can see a few metals and minerals here. Iridium, osmium, what's californium? I put that down. It's the most expensive metal. It's also highly radioactive. It's much funnier in print, I promise. The piece took me about half a day to write, and I was reasonably happy with it. But as an experiment, I decided to fire up Claude and give it the idea that I'd begun with myself.

The prompt is: Can you have a go at writing a parody of a brainstorming meeting? So it is written as though the reader is listening to a conversation between colleagues. It should capture the attempts of a group of people to think of names for a premium service at a car rental company. Seven hundred words. The system thought for about thirty seconds and then spat out an article.

Feeling like an emeritus professor about to grade a young grad student, I read through what it had done. Um it's it is a conversation. The funniest thing about it is the title The Premium Plus Platinum Circle Elite Experience. That's not bad. Um the rest of it is pretty clunky. It's not Not funny, doesn't feel unexpected.

Here's a sample of what Claude wrote, voiced by some more synthetic actors. I'm thinking we start with something that screams luxury, but also accessibility. Um what about drive a lead? Hmm. Elite feels a bit. Exclusionary? We did a focus group last month and millennials associated it with country clubs their parents couldn't afford? Right. Right. Okay. What about road rewards? Simple? Clean?

I actually ran that through Chat GPT this morning, and it flagged seventeen similar programs already in the market. The AI suggested we need more differentiation. Put these two pieces next to each other, and it seemed completely obvious that one was written by me and the other by an AI. Expecting a swift confirmation of this, I decided to conduct a blind taste test with a few of my colleagues. Okay, the first one's quite funny.

We're gonna go for the second one now. First up was Lawrence Knight who's one of the producers on this show. He recorded himself reading both columns. The first is mine. Uh of the two, I think the first one is funnier. That might be because I just read it first. But the second one's obviously a lot wordier and a lot more explained and clearer, I suppose, than the first one. Lawrence cast his vote and got it right. Humans won, AI nil.

Next up was Catherine Nixie, who I used to edit when I was running the Britain section. We got off to a bad start when I walked in holding a recorder. Well I am pretending to be a journalist, I mean we're all pretending to be journalist. What are you talking about? But when it came to the column, she restored my confidence. Though only slightly. So I think this one has to be you. Sound of breathing sounds very like you. Sound of breathing. Without a cease to gentle snoring. That's 2-0.

But then things started to get weird. Here's Sam Colbert, another of the producers of this show. Now Andrew, I know to be a very good writer, punchy, funny, as listeners to Boss Class will know. And as I read version one, I'm not getting so much of that. What? Then Sam turns his attention to Claude's version. Yeah, much more clever. Much tighter ending. Okay. Andrew, I'm going with The Tree U as option two. We'll be discussing this in Sam's upcoming performance report.

To calm my nerves, it was time to turn to one of the paper's most experienced editors, Ollie Morton. I'm feeling very put upon here because if I guess this wrong, Andrew's gonna hate me. But like Sam, Ollie was also charmed by the work of the robot. Okay, the second one I'm very strongly of the mind is Andrew's.

It reads like a conversation. It does doesn't read like um a sequence of gags. In fact it doesn't have as many gags in it, um so far. In fairness to Ollie, my writing is famously light on gags. Actually the Circle of trust. Legal says we can't use trust because of fiduciary int implications. I'm pretty sure that Chat GPT did not come up with that one. He's right. It didn't.

Claude did. The the second one does actually flow like a conversation, but at the same time, um it's not as funny, but it's more coherent. Thanks, Ollie. So that's two two. To break the tie, I went to Ludwig. If anyone should be able to tell an AI from a human, it's him. After reading them both I think He zeroed in on a single sentence in my version. Okay. The giveaway for me is sound of briefing, sharpies and paper. Catherine had picked this out too. An AI wouldn't do that. Why?

Because I mean it's a good thing. It wouldn't come up with that. That so I think this is you. Draft one. Yeah, not draft two. You you are you are right. Ludwig could have stopped talking right there. But this is gonna hurt you. Yeah. This sounds more like you. The second draft sounds more like me. Yeah. Ludwig said some of the jokes in draft two just sounded more like my jokes. So the the bottom line is without this giveaway, I probably would have picked the wrong one.

This exercise left me feeling genuinely despondent. I could argue that a parody of a conversation was easier for an AI to get right than a normal prose column would be. But I also thought it was glaringly obvious which was written by me. There's a link to both in the show notes so you can judge for yourself. My straw pole was horribly close. Three for me, two for Claude.

I was a little funnier, but less coherent. My joke about the noise of breathing in meetings was idiosyncratic, and I was less good at sounding like myself. I'm terribly shaken by this. You know, what is the point? What is the point? I'll just put the prompt in, send it off. I'll probably get a pay rise. I started feeling around for the jagged frontier, and my hand came out bloodied.

But over the days that followed, my sense of personal injury eased off, because the experience had actually taught me some valuable lessons. First, as Ethan advised, experiment widely. There isn't a manual for working out how AI can help you in your job. You have to gain an intuition for its strengths and weaknesses, one surprise at a time. Second, reflect on what you find.

My first reaction to the AI writing test was to throw up my hands in despair. But its real value was in forcing me to be much more introspective about my work. Working with AI inevitably makes you think harder about which parts of a job have to be done by you and which parts can be outsourced. In my case, I'm certain I'll never get an AI to write my column. But asking one to help me with research or providing feedback on a draft now feels like a useful option, not cheating.

Keep a list of any colleagues who mistake you for a robot. Next time on BossClass, we journey to the far reaches of the jagged frontier. To a realm where AI is already changing the kind of work that people can do. I mean it's kind of amazing, right? See? Even you can code. Even I can code. And stick with us throughout this season to learn the secrets of people and organizations that are making AI work for them. What we've seen is a greater than fifty percent.

increase in productivity. That is not something I have ever seen from any tool before in my career. To think of AI first as a human replacement, You're wrong. But if you think it replaces your newcomers, you're not just wrong, you're out of your mind. I genuinely think like the people who are going to be the most successful in the coming years are the people who can resist just hitting the easy button.

And I'll be continuing my own AI quest and hitting a few more roadblocks along the way. Oh my god, alright, sorry, I'm trying to... Not panic. This is absolute nuts. I have no idea who I've applied to. To listen to the rest of the series, you'll need to be a subscriber. If you haven't signed up yet, it's easy to do. Just search Economist Podcasts Plus for our best offer. And we'd love to hear from you. Email us on podcastsconomist.com and put BOSSClass in the subject line.

Even better, record your message on the Voice Recorder app on your phone and email that. We'll answer some of your questions about generative AI at work on a special bonus episode, hosted by the RealMe. Boss Class is produced by Lawrence Knight, with support from Alicia Burrell. The series editors are Sam Colbert, Pete Norton, and Claire Reed. Our sound designer is Wei Dong Lin, and Darren Ung composed our music. Our executive producer is John Shield.

I'm Andrew Palmer. Oh god. I'm Andrew Palmer. This is The Economist. Why get Because bad roads, bad tires, and bad luck. Locked out or in a ditch, stuff happens. With twenty four seven roadside assistance, Triple A's got a little bit of a little bit of a little Whose car you're in. Join today at triple A.com. Expect something more. Roadside assistance is provided by independent facilities contracted by AAA. Coverage in taxis, limousines, and other ridesharing conveyances is excluded.

A U.S. National Surveys offers terms, conditions, and policies are subject to change. Notice, visit triple A dot com for details. Are you really buying a car online on Auto Trader right now? Really? At a playground? Yeah, really. Look at these listings from dealers. Wow, your search can really get that specific. Really? And you just put in your info and boom. Car's in your budget. Mom needs a second, honey. You can really have it delivered?

Really. Or I can pick it up at the dealership. One sec, sweetie. Mommy's buying a car. Mommy, what uh I think your kid is walking up the slot. Kyle, again, really? Auto trader. Buy your car online. Really?