The Intelligence Squared Economic Outlook, with Tim Harford, Part Two
31:00 Share

This episode is sponsored by Oracle. Even if you think it's a bit overhyped, AI is suddenly everywhere. From self-driving cars to molecular medicine to business efficiency. If it's not in your industry yet, it's coming fast. But AI needs a lot of speed and computing power, so how do you compete without costs spiraling out of control? Time to upgrade to the next generation of the cloud, Oracle Cloud Infrastructure, or OCI.

OCI is a blazing fast and secure platform for your infrastructure, database and application development, plus all your AI and machine learning workloads. OCI costs 50% less for compute and 80% less for networking. So you're saving a pile of money. Thousands of businesses have already upgraded to OCI, including MGM Resorts, Specialized Bikes and Fireworks AI.

Right now, Oracle is offering to cut your current cloud bill in half if you move to OCI for new U.S. customers with minimum financial commitment. Offer ends 31st of December 2024. See if your company qualifies for this special offer at oracle.com. That's oracle.com.

Bada, bada boom, sold. Huh? Just sold my car on Carvana. Dropping it off and getting paid today. Already? What, you still haven't sold yours? You told me about it months ago. I just... Is the offer good? Oh, the offer's great. Don't have another car yet? I could trade it in for this car I love. Come on, what are we waiting for? Ah, you're right. Let's go.

Whether you're looking to sell your car right now or just whenever feels right, go to Carvana.com and sell your car the convenient way. Terms and conditions apply.

This episode is brought to you by GlobalX. Since 2008, GlobalX ETFs has been committed to empowering investors with unexplored intelligence solutions. GlobalX specializes in exchange-traded funds that offer exposure to the artificial intelligence ecosystem, including themes like data centers, robotics, semiconductors, and cloud computing. To learn more about GlobalX's entire suite of ETFs, from covered calls, fixed income, emerging markets, and more, visit globalxetfs.com.

Welcome to Intelligence Squared, where great minds meet. I'm producer Mia Cirenti. For this episode, we're rejoining for part two of our conversation with the economist Tim Harford. Tim is a senior economist for the Financial Times, where as the undercover economist, he reveals the economic ideas behind everyday experiences. His recent book is How to Make the World Add Up: Ten Rules for Thinking Differently About Numbers.

This is the second part of our two-part episode, The Intelligence Squared Economic Outlook, in partnership with Guinness Global Investors. Guinness Global Investors is an independent British fund manager that helps both individuals and institutions harness the drivers of future growth to achieve their investment goals. If you haven't heard part one, do just jump back an episode and get up to speed. Now it's time to rejoin the conversation recorded recently at London's Tambrackle in Uptickhead.

Here's our host for the evening, BBC News presenter and royal correspondent, Jonny Diamond. Can we talk about the opinion polls? Did you believe them? Well, they said it was close, didn't they? I mean, I believed that it was close. But I thought they'd probably be wrong one way or another. I didn't know how much.

I didn't know in which direction. So it's a bit strange. So the last three presidential elections, the polls have all been wrong in favor of the Republicans. So Biden had such a big lead in 2000 that it didn't matter. But it turns out that that election was tighter than it seemed and his lead was not as big as it seemed. Clinton had a lead over Trump in 2016. Kamala Harris had a lead in the popular vote in

that in the opinion polls that turned out not to materialize. So that's three successive presidential elections in which the polls have kind of suggested the Republicans are doing worse than they end up doing. On the other hand, the midterms two years ago, Democrats did better than the polls suggested. So it's just difficult to be an opinion pollster. And I think we need to recognize we treat opinion polls as though...

as though this is a simple thing to do, right? You sort of phone some people up and then they tell you who they're going to vote for and you kind of write it down. Talk to 500 people and you'll probably be right. Talk to 2,000 people and you'll definitely be right. That's not how it works. But the first thing to bear in mind is like, you call people and you might, I don't know what the precise numbers are these days, but you might call 20 people and get one person to answer your question. So what do those 19 people think?

We know they're different from the one person who answered your question, because they wouldn't answer your question. Or maybe they weren't in, but they were different from the person who was in. They've got different habits. They were available at different times. Or maybe they said they wouldn't answer the question. That immediately makes them different. So you're trying to infer...

from a sample of 5% of the population or whatever it is, who you know are systematically different from the other 95%. What are you going to do with that? You can make all kinds of corrections, but a lot of the corrections are just, "Well, how wrong were we last time and in what direction? Should we correct like that or is that overcorrecting?" It's just a really, really difficult job. And fundamentally, if you took a step back, the opinion polls are basically right.

They said it was about 50-50, it was about 50-50. It's just that we actually want to know who's going to win the election, so we're not happy with about 50-50. We actually want to know. But if it was 70-30, and it turned out Arne actually was 68-32, nobody would care. It's the fact that they keep being wrong enough to change the outcome, or not to change the outcome, but to have misled us about the outcome that bothers people. But we shouldn't really be surprised. And I don't see...

any change in polling technology that's going to fundamentally change the problem.

The number one cloud ERP brings accounting, financial management, inventory and HR into one fluid platform. With one unified business management suite, there's one source of truth giving you the visibility and control you need to make quick decisions. With real-time insights and forecasting, you're peering into the future with actionable data.

If I had needed this product, it's what I'd use. NetSuite by Oracle is just a really smart tool for businesses looking to make proactive moves and make the best out of future opportunities. Whether your company is earning millions or even hundreds of millions, NetSuite helps you respond to immediate challenges and seize your biggest opportunities.

And speaking of opportunity, download the CFO's Guide to AI and Machine Learning at netsuite.com. This guide is free to you at netsuite.com. That's netsuite.com.

This episode is brought to you by AWS. Amazon Q Business is the new generative AI assistant from AWS. Many tasks can make business slow, like wading through mud. Help! Luckily, there's a faster, easier, less messy choice. Amazon Q can securely understand your business data to help you streamline tasks, like summarizing quarterly results or doing complex analyses in no time. Q got this. Learn what Amazon Q Business can do for you at aws.com slash learn more.

Hey, aren't you that PBM? Middleman. At your service, doctor. Don't you get rebates that save money on medicines? Oh, PBMs like me get big rebates. So why do patients tell me they're worried about their costs? No one says we have to share the savings with patients. Congress should make sure medicine savings go directly to patients, not middlemen. Visit prma.org slash middlemen to learn more. Paid for by Pharma.

You've also written about what I think you called the forecasting paradox, that accurate forecasts are not necessarily useful and useful forecasts are not necessarily accurate. I hope I've summarised that correctly. Yes. Is there any point in forecasts? I think there is a point in forecasts, but you need to understand what they're for. So let me give you a couple of examples of forecasts.

I think the weirdness. I give you endless examples of bad forecasts, but let's think about just the demand for forecasts. So here's one. You might remember about eight years ago now, I think it was Leicester City won the premiership. And that year...

The BBC's chief football writer, I think it was Phil McNulty, could be wrong. Anyway, big football expert at the BBC. At the beginning of the season, he said, "I think Chelsea are going to be top of the table by Christmas. I think Leicester are fighting for relegation. Well, come Christmas, Chelsea are in the bottom five, Mourinho's been sacked, the Chelsea manager, and Leicester are at the top."

So he says, "Well, that's what I said at the beginning of the season. Here we are. We've got halfway, and I was completely wrong. Fair enough."

Anyway, I thought I'd make some forecasts for the rest of the season. Why does anybody care? Why does anybody care what you think is going to happen, given what you thought was going to happen has manifestly failed to happen? And yet, I read the article, and I thought, "Oh, interesting, very insightful." Pringles, it's like the snack food of the intellectual world. We can't help but consume them, although we know they're no good.

So that's one thing that's going on. There's this insatiable demand for forecasts. And I think one of the reasons why there's an insatiable demand for forecasts is because we're all a bit busy to really get into the details of anything. So if you, you know...

What's going to happen in Afghanistan? You sit down with an expert on Afghanistan to explain the government the Taliban have set up and the different foreign affairs pressures. But no one's got time for that. Say you ask for a little forecast, and the expert says, well, I think things are going to be terrible for the next three years, or I think things are going to really pick up in the next three years. You go, oh, great. And you come away with the illusion you've understood something about Afghanistan, but you haven't understood anything. It's just some guy making stuff up.

And he or she could have told you something useful and interesting, but you didn't have time. So you just asked for the forecast. So that's one thing. Second thing related to the demand for forecasts. I was speaking at an event a little bit like this one, but the person interviewing me was less good-looking and the audience were less insightful. And somebody said, this is the first...

event that we've had since late 2019. And in late 2019, we had an eminent scientist come to us and he said that he was really worried about pandemics. He showed us how air travel could spread viruses and a pandemic could go all around the world before we could track it and it could shut down supply chains and it would kill tens of millions of people. So the questioner added or asked me, could you

"Give us a forecast that's as good as that one." And then I thought, "Hang on a minute. You who were in that room, who heard that forecast in October 2019, you were privileged to have one of the best, most relevant, most consequential, most accurate forecasts anybody's ever received. What did you do?"

And the answer was, oh, I didn't do anything. I mean, maybe I went home for dinner and I told my spouse, oh, there was this guy at this conference who was talking about pandemics. It's all very interesting. Nobody did anything. Also, what were you supposed to do? I mean, the big scientist comes and says he's worried about pandemics, but he doesn't know when. I mean, as it happened, in about six weeks. But he didn't know that. And none of the people in the room were in a position to do anything about it. So it was, I mean, it was an accurate forecast. It was an incredibly accurate forecast. But

It's just cheap talk. So all of this is kind of by way of preamble. What is a forecast for? Generally, it's for cheap intellectual gratification. It's not for anything. And that's true whether or not the forecast is true or false. It doesn't help us. But forecasts can help us.

I'll give you a couple of quick examples. So, a long time ago, I used to work in scenario planning. And the whole idea of scenario planning is you get people in a room who really approach a problem from all kinds of different directions, not just the economists, you've got the political scientists, you've got the environmentalists, you've got the business types.

and they discuss what's going on in a particular situation, and then you start telling stories about what might happen. And the amazing thing about stories is that people remember stories, they really get to grips with stories. And the stories, you can hang lots of detail, all that stuff you couldn't be bothered to find out about Afghanistan. Suddenly, you're really paying attention to the story. But here's the thing: you tell two stories, and they're mutually contradictory. So now you can remember both stories, because they're good stories, they're memorable stories,

You can talk about the different stories, you can discuss your strategy in the light of the different stories. Well, that might work in this world, but what about this world? And you're not going to forget the fact that you don't actually know what's going to happen because you just heard two perfectly believable stories that made perfect sense to you and were very persuasive, but can't both be true. So that's one way of approaching the problem of futurology.

Just another quick observation, then I'll shut up, is Phil Tetlock, who's most famous for his work on super forecasting. He gets lots of people in a room and... Not in a room, he gets lots of people online to answer questions and finally calibrate their forecast, and they come back and they check whether the forecast were right. And some people are very, very good at this.

So he runs these forecasting tournaments. One of the things he's found is that participating in a forecasting tournament for three months is very good for your intellectual hygiene. You start seeing shades of grey where other people see black and white. You moderate your political opinions. You think about people on the other side of the political spectrum in a more empathetic way. You start to understand where they're coming from. You're less quick to judge them.

And this is all because if you're trying to work out what's going to happen, you have to actually start taking reality seriously.

So forecasting done in the right way, maybe in a forecasting tournament, an ongoing process, or maybe in a scenario planning workshop, can really help people think through the complexities of the world and make better decisions. And that is true whether or not the forecast is accurate, which is a good thing, because you know the thing about accurate forecasts, you don't know they're accurate until it's too late.

Do you use forecasts? I mean, apart from as intellectual playthings, do you use them, do you trust them? No, no, I don't trust them at all. Unless you've got a very, very high data situation. So if, you know, if I'm... I mean, I don't actually make these sort of decisions, but say Amazon were to come to my publishers and say, we're thinking of cutting the price of...

Tim's book by 10% and we think that if we cut the price of Tim's book by 10%, that's going to, maybe I get a 10% margin, or 20% margin, so it's going to halve that margin. So you're making half as much money per book, but we think we're going to sell three times as many books if we cut the price by 10%.

I basically believe that kind of forecast because they've got so much data. I mean, they might be lying to me, but I believe that they know the answer to that question with fairly good accuracy because they've got an enormous amount of data. You know, if Google tells me it's going to take, you know, an hour to get somewhere, generally, that's roughly right. I mean, as Johnny can tell you, I believed Google how long it was going to take me to get here, and it was slightly wrong. He was very late. I was just trying to, you know...

Keep Johnny awake, really. But those are very high data forecasts. They're sometimes wrong, but they're not often wrong by a lot. So if there's enough data and something's repeated, then sure, I believe the forecast. But normally when we're talking about forecasts, we're not usually talking about that. We're talking about these big picture geopolitical things. And there's so much hot air.

So much lazy thinking, both by the people producing the forecast, but also just as importantly by the people consuming them. I have to say, I don't take them very seriously. How about listening to the sounds of Istanbul? Beautiful, isn't it? But you can't discover the coolest city in the world just by listening. Check istanbul.goturkiye.com now and plan your Istanbul trip today.

This episode is brought to you by Shopify. Upgrade your business with Shopify, home of the number one checkout on the planet. ShopPay boosts conversions up to 50%, meaning fewer cards going abandoned and more sales going cha-ching. So if you're into growing your business, get a commerce platform that's ready to sell wherever your customers are. Visit Shopify.com to upgrade your selling today.

Let's move from forecasting to prophesying. And let's talk a little bit, if we could, about the big developments in the economy and things that are shaping it. The buzz is all around AI and how it will change things. What in particular do you think we can expect from AI development in the next 18 months or so and how that will apply and how that will change the economy?

So, the first thing to say is it would be great if it would change the economy because, as we discussed earlier, we've had terribly slow productivity growth. We're kind of waiting for the robot takeover, please, at some stage, be nice. If this stuff is going to transform the economy, why are we not seeing it in the productivity statistics, which is a question that economists have been asking since about 1986.

I mean, what am I expecting AI to do? I don't know. One of the interesting things is that we don't really know what it can do right now, even if there's no further advancement in AI's capabilities from where it currently is. There's all sorts of stuff to explore about how humans and AI's can best work with each other, how we can figure out error correction, how we can stop it making stuff up.

That's, I think, enormously fertile. Just exploiting the stuff that, in principle, we could already do, but we just need to sort of figure out how to get organized. That, I think, is fascinating. And the... So, Ethan Mollick, who writes very interestingly about AI, he talks about the jagged frontier of AI capabilities. So, the...

you ask an AI to do something and it's very hard to predict whether it will be able to do it or not. There's a wonderful online test, sadly I've blanked on the person who created it, he's a Google engineer, but you can go on this test and you can...

Do you think AI can prove Pythagoras' theorem or whatever? Do you think AI will be able to tell you which is bigger, Paris or Tokyo? Do you think AI can win a game of Noughts and Crosses? To which the answer is no. Do you think AI can set up a website that will run a Noughts and Crosses game

and the website will play perfectly, the answer is yes. The AI can't beat you at noughts and crosses, but it can set up a website that will beat you at noughts and crosses. All this sort of stuff is so hard to predict what it's going to do. I went through, and I was basically 50/50 as I went through it. I've got no idea of the capabilities of AI. I just don't know. So there's a lot of exploration going on.

So that's interesting, it's exciting. I mean, lots and lots of huge opportunities, lots of risks. I'm not worried about the robot takeover just yet, but I'm worried about all the ways in which it can just be used in a very lazy way or can create discrimination or it's full of errors. One thing that I always emphasize when thinking about this is we tend to think of the AI is going to take our jobs

Have you heard the Google large language, Google notebook language model, Notebook LM? So it's designed to help people

It's helped writers, basically. So, you can grab PDFs that you're reading, your own work, stuff that you're researching, and you can ask the notebook to -- you can ask the AI to identify trends of what you found or find some unexpected connections, and it will footnote everything. It will say, "This is where I found this fact." It's not just making stuff up. So, it's really interesting.

Then they added this feature. So they had this for about a year. People were talking about it. Well, that seems interesting. Then they added this feature where you can just go, make me a podcast. So you can upload a chapter of my book. Well, hopefully you can't because hopefully you don't have the unencrypted version. But I can upload a chapter of my book and just press a button. And these two rather giddy Americans...

we'll start talking about how this chapter is completely amazing and you've never heard anything like it. And they're totally artificial voices, it's a totally artificial conversation, but it really feels like they both read the chapter in detail, got it, and could talk in an intelligent way. So we are totally screwed, you and I, Johnny, we are in big trouble. Anyway, so I interrupted myself, which the computer would never do, they'd be doing a much better job than I would.

We talk about the computers taking our jobs, we talk about the robot taking our jobs, the AI taking our jobs. That's not what happens. What happens is they take tasks, they do particular tasks. And we, because we're very adaptable, we kind of contort ourselves around the new technological capability. So you can see this most obviously when they introduced

lasers and barcode scanners at supermarkets. It took a little while to think about it, and then suddenly we're all working as unpaid checkout assistants. Because the AI can't do that for us, but it can provide us with the tools and we sort of fit in. So they reshape jobs. They make jobs different. They'll take particular capabilities rather than taking whole jobs. And I think that's the way to think.

to think about what AI is likely to do. And just to give you two very, very quick examples of this, one hopeful and one less hopeful. So digital spreadsheets. So the Lotus 1-2-3, actually before Lotus 1-2-3, VisiCalc. VisiCalc's the first digital spreadsheet, 1979. It just transforms accountancy overnight. Suddenly, accountants get hold of this thing, and they don't tell their clients. So their clients say, we need a rush job on the accounts. And the accountant's like, yep.

And they'll just leave it for two days. And then they'll send the bill for two days' work until the clients figure it out. So...

You would think, "Well, that's going to put the accountants out of work. What is Microsoft Excel but a robot accountant? I mean, that is what a robot accountant is." But it doesn't put the accountants out of work. There are more accountants now than ever. They're just doing more stuff. They're doing more creative stuff. Who doesn't want a creative accountant, right? So this job that-- There used to be these people called accounting clerks or accounting clerks who would do the arithmetic. And now the computer does the arithmetic, but you still have accountants and they're doing more interesting things. That's a hopeful vision for AI.

Here's a really depressing one. The Jennifer unit. So the Jennifer unit's a bit like this, only it's got an earpiece. And if you're working in a warehouse, say, as an Amazon picker, or you're just sort of wandering around Ocado or whatever, you're picking stuff off shelves. And it used to be you had to know where stuff was...

and you'd go to the shelves, you'd get the stuff off the shelves, and you could tick it off a list, and that's how you'd get all the stuff. But now you have the Jennifer unit. The Jennifer unit knows where everything is. So the Jennifer unit will just say, "Okay, take 20 steps forward to aisle B, then go to shelf 6."

then take 13 copies of Tim Harford's "How to Make the World That Up." Actually, it won't say that. It won't even say that. It'll say, "Take five copies of Tim Harford's "How to Make the World That Up." "Take five more copies of Tim Harford's "How to Make the World That Up."

take three more copies, because it doesn't trust you to count to 13. You can't count to 13, that's complicated, people get it wrong. So it'll do it in chunks of five. All it wants is people's eyes and their hands. It's treating them like flesh robots.

In the case of Excel, you have a job that's already quite interesting being made more interesting because the robot, the spreadsheet does the boring bit. In the case of the Jennifer unit, you have a job that's already quite boring being a warehouse picker, and the little bit of intellectual stimulation you had is gone, and now you're a flesh robot, and people who work in warehouses go to sleep dreaming of Jennifer talking to them.

So there's different ways in which AI can play out. It's going to play out differently for different jobs. Some of it's going to be very unpleasant and some of it could be really cool. Right, if there's one phrase I will take away from this evening, it's the phrase "flesh robot". Thank you for that, Tim. I am bitterly sorry, but we do have to wrap it up there. We are out of time. And I am sorry if you had further questions for Tim, but the time has come, we have to wrap it. I need to thank Tim.

I've not heard so much laughter in a discussion about economics. They were laughing with you, I promise. It's not at you. Thank you for coming as well. Thank you to Guinness Global Investors for being part of this event and to Intelligence Squared. But most of all, thank you very much to the audience. It's been brilliant to have you here. We'll leave it there. Thank you and good night. Thanks for listening to Intelligence Squared. This episode was created in partnership with Guinness Global Investors.

It was produced by myself, Mia Sorrenti, and it was edited by Bea Duncan. For more information on Guinness, just head to guinnessgi.com or see the link in the episode description. You've been listening to Intelligence Word. Thanks for joining us.

UTIs are the worst. I've been there. One year, I had eight UTIs. If you get UTIs, then you understand how awful the cycle can be. I was taking all the precautions. And cranberry products? They just never worked for me. I was desperate for a way to be proactive.

It was hard on me and on my husband. It was tough to see her in pain, and I wanted to help. I'm Jenna. And I'm Spencer. With Spencer's background in biochemistry and our shared frustration when it came to UTIs, we were inspired to start Eucora. At Eucora, we make innovative urinary tract supplements and UTI relief products.

Our effective urinary tract supplements finally give you a way to be proactive. Feel like you've tried everything? We get it. We have a money-back guarantee so you can try risk-free. If you're not happy, you'll get a full refund. We're on a mission to help women get their lives back. Ready to join them? Get 30% off Eucora when you go to eucora.com slash audio. That's U-Q-O-R-A dot com slash audio. Marketing is hard.

But I'll tell you a little secret. It doesn't have to be. Let me point something out. You're listening to a podcast right now, and it's great. You love the host. You seek it out and download it. You listen to it while driving, working out, cooking, even going to the bathroom. Podcasts are a pretty close companion.

And this is a podcast ad. Did I get your attention? You can reach great listeners like yourself with podcast advertising from Libsyn Ads. Choose from hundreds of top podcasts offering host endorsements or run a pre-produced ad like this one across thousands of shows to reach your target audience in their favorite podcasts with Libsyn Ads. Go to LibsynAds.com. That's L-I-B-S-Y-N ads.com today.