Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
The Macro Brief – The good, the bad and the ugly of AI
27 Plays1 year ago
James Pomeroy, Global Economist, discusses the likely impact of artificial intelligence - including what it could mean for productivity, jobs, and emerging markets.
Disclaimer: https://www.research.hsbc.com/R/101/nXPrpkg Stay connected and access free to view reports and videos from HSBC Global Research follow us on LinkedIn https://www.linkedin.com/feed/hashtag/hsbcresearch/ or click here: https://www.gbm.hsbc.com/insights/global-research.
Hosted on Acast. See acast.com/privacy for more information.
Recommended
Transcript
Introduction and Subscription Details
00:00:02
Speaker
Welcome to HSBC Global Viewpoint, the podcast series that brings together business leaders and industry experts to explore the latest global insights, trends, and opportunities.
00:00:13
Speaker
Make sure you're subscribed to stay up to date with new episodes.
00:00:16
Speaker
Thanks for listening.
00:00:17
Speaker
And now onto today's show.
00:00:24
Speaker
This podcast was recorded for publication on the 30th of May, 2024 by HSBC Global Research.
00:00:29
Speaker
All the disclosures and disclaimers associated with it must be viewed on the link attached to your media player.
00:00:33
Speaker
And don't forget that you can follow this weekly podcast on Apple and Spotify, or wherever you get your podcasts.
00:00:37
Speaker
Just search for The Macro Brief.
00:00:44
Speaker
Hello and welcome to the Macrobrief.
AI's Impact on Productivity and Society
00:00:45
Speaker
I'm your host, Piers Butler.
00:00:47
Speaker
Today we're digging deep into the world of artificial intelligence.
00:00:50
Speaker
Public interest in AI has boomed over the past year or so, but it's also led to a significant degree of polarization.
00:00:58
Speaker
Some are excited at AI's potential to drive productivity, while others worry about potential side effects, such as jobs being displaced.
00:01:06
Speaker
So where do we stand now?
00:01:07
Speaker
And what could AI mean for the global economy?
00:01:10
Speaker
Well, last week, our economics, data science and ESG teams came together to publish a major report on the good, the bad and the ugly side of artificial intelligence.
00:01:19
Speaker
I'm joined by one of the report's lead authors, global economist James Pomeroy.
00:01:24
Speaker
James, welcome to the podcast.
00:01:25
Speaker
Thank you for having me.
00:01:27
Speaker
So James, the irony for me is that AI has actually been used productively in the real economy for several years.
00:01:32
Speaker
So just remind me as to why the arrival of the ChatGPT has been such a game changer.
00:01:38
Speaker
So as you say, we've been using AI in a whole number of tools over the course of the last few years.
00:01:43
Speaker
but probably in ways that we don't realise.
00:01:44
Speaker
It's hidden within infrastructure or within interfaces rather than us directly having access to what the AI output is.
00:01:51
Speaker
And that's one of the areas where things like chat and GPT have changed because people can now see, I put this in, there is the output in terms of AI, and it's made people realise actually a lot of the power of these tools.
00:02:01
Speaker
But beyond that, you're also seeing the capabilities of AI content development,
00:02:06
Speaker
go well beyond where it was previously.
00:02:08
Speaker
We've seen obviously the arrival of large language models which allow AI to do much more than they could before, but also things like image generation, audio generation and a whole range of different tools that can be used in a number of different ways throughout the economy.
00:02:21
Speaker
So AI has been here for some time, it's just the range and scale of applications is increasing incredibly quickly, which is why we've got to pay more and more attention to it.
00:02:31
Speaker
And in your report, you talk
Economic Impacts of AI Adoption
00:02:32
Speaker
about psychological shock for some.
00:02:34
Speaker
And also, as I said, the intro, polarisation.
00:02:37
Speaker
Can you expand on that?
00:02:38
Speaker
Yeah, it's amazing.
00:02:39
Speaker
Whenever you get a big breakthrough in something to do with technology, there's people who love it and people who hate it.
00:02:43
Speaker
And this has come out of, not out of the blue, but the speed of which we've seen AI spread into the public consciousness has been incredibly fast.
00:02:51
Speaker
And that's meant you've got a lot of people who are very excited about it and think, OK, this is going to be fantastic for the world and create loads of opportunities.
00:02:57
Speaker
And then you've got people who are just terrified by this and some of the potential side effects, as I'm sure we'll discuss.
00:03:03
Speaker
And that sort of pessimistic side of things can get really quite concerning, right?
00:03:07
Speaker
When you think about some of the potential downsides of AI on society and on the economy, potentially could be quite dramatic.
00:03:15
Speaker
So you can see why this polarisation has started to arrive over the course of the last year or so.
00:03:22
Speaker
OK, James, so let's get into it.
00:03:23
Speaker
Firstly, the good.
00:03:25
Speaker
Is it going to be a productivity miracle?
00:03:27
Speaker
It could well be.
00:03:29
Speaker
And now there's a lot of sort of caveats to how good AI is for productivity, and it comes down to firstly the speed at which we roll out AI tools.
00:03:37
Speaker
which we do generally expect to be relatively slow.
00:03:40
Speaker
There's a whole load of reasons why this is.
00:03:42
Speaker
It comes through from businesses having to be a little bit cautious because of not necessarily having the right training in place.
00:03:49
Speaker
Or it might be that they're a little bit conscious of some of the inaccuracies of NAI or some of the biases of NAI.
00:03:55
Speaker
That means that that rollout is a little bit slower.
00:03:56
Speaker
So that speed really matters in terms of how we're thinking about productivity.
00:04:01
Speaker
But on top of that, we need businesses to lean into it as well.
00:04:03
Speaker
And that goes back to that training tool, making sure that people are equipped to use AI tools effectively, but also thinking about how you can use those productivity benefits in other ways.
00:04:12
Speaker
And the example we've used in the report and we've used previously is saying AI is probably one of the main reasons why four-day weeks in many services jobs becomes quite applicable.
00:04:21
Speaker
Because if you think about the implications of AI, it automates tasks.
00:04:25
Speaker
And if a certain number of your tasks, which some research from OpenAI suggests about 15% of tasks can be automated, you very easily start working towards a 20% and therefore the four-day week.
00:04:36
Speaker
So I think that's an area where the impact does matter.
00:04:39
Speaker
But really, these numbers, you're talking about...
00:04:41
Speaker
somewhere between 0% and 1% per year productivity growth.
00:04:45
Speaker
Because we also have to remember, not every job in the economy is exposed to AI and can see productivity benefits from it.
00:04:51
Speaker
So it's not like you're going to see a complete transformation of every single job that everyone's doing in the economy.
00:04:57
Speaker
But those areas of the economy where there is more exposure could well see some substantial productivity gains.
00:05:03
Speaker
And would it be fair to
Global Disparities in AI Integration
00:05:04
Speaker
say that ultimately productivity improvements will become generalized?
00:05:07
Speaker
And so in the short term, the debate is who are going to be the early adopters and who are going to get an advantage before it becomes more generalized?
00:05:15
Speaker
Are there signs that some countries are earlier in the adoption?
00:05:18
Speaker
There is some sense.
00:05:19
Speaker
So typically you'd think, OK, who are the biggest winners from this in the data?
00:05:22
Speaker
Well, you'd look at the exports coming out of Taiwan, for example, in terms of the investment happening across the world.
00:05:27
Speaker
And you say, well, there's a pretty big winner straight off the bat.
00:05:30
Speaker
And in looking at that sort of Taiwanese trade data over the last year, it's a complete game changer for thinking about how we analyze the Taiwanese economy.
00:05:38
Speaker
But equally, you can look at the economies where you're seeing more of this technology being rolled across businesses.
00:05:42
Speaker
And the US does stand out here.
00:05:44
Speaker
And a lot of these sort of AI readiness indices put the US near the top alongside some more sort of typically tech savvy economies in Europe.
00:05:52
Speaker
So you're looking at those are the sort of economies who are best placed to benefit.
00:05:56
Speaker
But you also have to think about this through the nature of work.
00:05:58
Speaker
So in a lot of economies where you have a lot of manufacturing jobs, a lot of agricultural jobs in the emerging world, the benefits of AI on a lot of those jobs is relatively small because AI impacts services jobs.
00:06:09
Speaker
And so those economies where you've got a lot of higher skilled services jobs, they're the ones where it's easier to have productivity generated
00:06:16
Speaker
benefits, and a lot of analysis suggests that a lot of the jobs in those economies are complementary to AI.
00:06:22
Speaker
So they're improved by working with AI rather than necessarily being at risk.
00:06:26
Speaker
So it's not so simple just to say these are the countries that win, these are the sectors that win, when in reality there is a much more nuanced approach here on a sort of case-by-case basis.
00:06:36
Speaker
Yes.
00:06:36
Speaker
And in fact, we did an interview with your colleague, Mark McDonald, where he did that very experiment where he set a task for one of his team and set it for Chad GPT.
00:06:45
Speaker
And the conclusion was that the combination of the two had ended up being a much better result.
00:06:50
Speaker
That's it.
00:06:50
Speaker
And it's going to be a lot of jobs like that.
00:06:52
Speaker
I mean, you think about the job as an economist, it's going to be one that's going to benefit from working with AI tools.
00:06:58
Speaker
AI will allow us to do some of our job quicker, more effectively, and arguably better.
00:07:02
Speaker
And that will free up more time to do the things that AI isn't so good at doing.
00:07:05
Speaker
And that could be interacting with clients, doing presentations, coming up with ideas.
00:07:09
Speaker
And that's where I think you're going to see a lot of that interaction.
00:07:12
Speaker
But we always do have to remember that AI doesn't impact everybody necessarily.
00:07:15
Speaker
Now, there's a lot of jobs out there that will see very little impact from the spread of AI through the economy.
00:07:21
Speaker
And that's typically on the physical side of things, where a lot of jobs may have some impact indirectly.
00:07:27
Speaker
But in reality, there's still a lot of jobs out there on the physical side of the economy that won't see any impact to start with.
00:07:33
Speaker
OK, so let's turn to the bad.
00:07:35
Speaker
I guess there is this age-old dilemma of capital versus labour.
00:07:39
Speaker
Yeah, it's going to be a big question because one of the first things that people ask me when they talk about AI is about job losses.
00:07:46
Speaker
And essentially, everyone says AI is going to take everyone's job and we're going to see all these job losses.
00:07:50
Speaker
And we mentioned the productivity improvements.
00:07:52
Speaker
Well, if you don't need as many people because everyone's so much more productive, how many jobs are going to get lost?
00:07:57
Speaker
And then people start to worry, OK, then what you see is this additional challenge that...
00:08:01
Speaker
You've got some people who have been displaced from their work and they're competing for those jobs that aren't affected by AI and that bids down their wages and you get these potentially problematic issues.
00:08:10
Speaker
So the big question here about can we firstly ensure that that friction of unemployment doesn't happen?
00:08:16
Speaker
But there's also a question of can we think about the training that people are going to need on the job?
00:08:21
Speaker
Now, when you start working with AI as well, AI's impact on the labour market isn't just about losing jobs.
00:08:26
Speaker
It's about making sure people can work with the tools correctly.
00:08:30
Speaker
But then there's obviously this other question of a lot of businesses who are going to benefit enormously from these AI rollouts, this productivity increase, does that manifest itself in just higher profits or does it manifest itself in higher wages?
00:08:43
Speaker
And that's where one of the big challenges
Potential Risks and Overinvestment Concerns
00:08:45
Speaker
is going to come.
00:08:46
Speaker
you could well see that a lot of these automation tools, including AI, lead us down a path of companies making greater profits at the expense of workers.
00:08:55
Speaker
And if that is something that comes through, then that, of course, has big distributional consequences too.
00:09:00
Speaker
And one final point that we raised in the report about some of this is thinking about a lot of the investment that's going into AI at the moment is potentially exciting, and it is exciting for many people.
00:09:10
Speaker
But some of that is essentially bringing forward that investment.
00:09:13
Speaker
And so there's people who are potentially going to really benefit, but actually may lose out in terms of the investments they're making today.
00:09:19
Speaker
So you could see AI become this hugely important thing through the global economy.
00:09:24
Speaker
But because of overinvestment today, a lot of that, a lot of those returns won't necessarily be as good as maybe people are hoping.
00:09:31
Speaker
And there is the point that you make in your report that you may find firms over-competing for the benefits of AI, and ultimately it's a consumer that may benefit.
00:09:40
Speaker
So because of over-investment and over-competition, companies may not necessarily reap the rewards to their bottom line.
00:09:47
Speaker
Exactly.
00:09:48
Speaker
That's sort of the story that played out in the 90s with the so-called dot-com boom, where you had a whole flood of investment.
00:09:55
Speaker
Loads of people thought they were making loads of money investing in this booming industry.
00:09:59
Speaker
And some people did make loads of money for a short period of time.
00:10:01
Speaker
Then they lost loads of money.
00:10:03
Speaker
But also you had that the underlying story was still a good one for consumers.
00:10:07
Speaker
Consumers still benefited from a wider range of products, cheaper products, and from the dot-com boom.
00:10:12
Speaker
So there was a benefit that accrued to the consumer, not necessarily to the investors or to the companies involved because they were engaged in very, very competitive industries here.
00:10:23
Speaker
So you had this sort of over-investment, brought forward a lot of growth.
00:10:26
Speaker
Everyone got very excited.
00:10:28
Speaker
But actually, the benefits to those companies or those investors never really accrued over the medium term.
00:10:33
Speaker
And the productivity increases were limited, you could argue, over the medium term as well.
00:10:38
Speaker
And there is a risk, of course, with AI that that's what happens.
00:10:40
Speaker
There's no sense of it that that's what's happening today.
00:10:43
Speaker
But there is a risk that we're essentially just bringing forward a little bit of investment that would have happened over a longer period of time into today.
00:10:50
Speaker
And the multiplier of that on the economy of the medium term is non-existent.
00:10:53
Speaker
And therefore, maybe there's a little bit of hype involved.
00:10:55
Speaker
And I think that's one of the big risks we have to keep in mind at the moment, that there's a huge possibility for AI to transform the global economy.
00:11:03
Speaker
But there is a sense of potentially
Regulation and Oversight of AI's Power Demands
00:11:05
Speaker
a little bit of hype starting to build.
00:11:07
Speaker
And it comes through the amount of questions we get on the subject almost everywhere we go.
00:11:11
Speaker
And maybe that's what we start to see in the coming years.
00:11:13
Speaker
We have to be careful if we do see a little bit of overinvestment in the space.
00:11:18
Speaker
James, it's probably fair to say that you and I are glass half full type of people, but we have to look at the ugly side of AI.
00:11:24
Speaker
And the first thing you talk about in the report is this whole question of the nature of truth.
00:11:29
Speaker
As an example, people could be listening to the podcasts and it could be completely fake.
00:11:34
Speaker
AI could have manufactured our voices and the whole discussion, and people could be none the wiser.
00:11:41
Speaker
I mean, a couple of years ago, that would have seemed completely fanciful, but it's no longer the case.
00:11:45
Speaker
So my question is, is the ugly side that we're not going to be able to tell what's genuine and what's fake?
00:11:51
Speaker
I think that's the biggest threat.
00:11:53
Speaker
And for all of the positives and all the opportunities that AI brings, this chance of creating essentially fake reality is a massive one.
00:12:02
Speaker
And it's hugely important when we think about financial crime, when we think about elections, when you think about news, all of these things that are so important to the global economy and to society are at risk in one way or another from the development of AI.
00:12:16
Speaker
And that's a huge challenge that we're not going to fix overnight.
00:12:19
Speaker
It's where regulation is going to be important.
00:12:20
Speaker
It's where oversight is going to be important.
00:12:22
Speaker
It's where training people to work alongside AI so we can almost get a sense of where things are not real.
00:12:29
Speaker
That's going to take time.
00:12:30
Speaker
And that's a huge, huge challenge.
00:12:31
Speaker
And say for all of the sort of economic cautious optimism that we have, this is something that we cannot get away from.
00:12:38
Speaker
And it poses enormous societal risks as well as potentially economic ones from this challenge that, as I say, isn't going to go away.
00:12:46
Speaker
In your report, you also highlight the environmental impact.
00:12:49
Speaker
With the growth of AI, there will be a lot of machines whirring away in the background to deal with the enormous computing power required by AI.
00:12:56
Speaker
What are your thoughts on that?
00:12:58
Speaker
Yeah, it's a huge challenge because it's not just about the computing power in terms of hardware.
00:13:02
Speaker
There's also the servers and the power that's needed to power all of those.
00:13:05
Speaker
And yes, there's some sort of solutions in some of these places, thinking about the sort of...
00:13:10
Speaker
upgrading of computer systems rather than just continually buying new ones and thinking about where you locate servers.
00:13:17
Speaker
So sort of in Scandinavia, where you have hydroelectric power and lower temperatures.
00:13:22
Speaker
But the power demands that you're seeing from using AI are astronomical.
00:13:25
Speaker
Then it means that for this to be sort of plausible and to get some of those productivity benefits,
00:13:30
Speaker
we're really going to need to see big breakthroughs in power supply.
00:13:34
Speaker
And either that's going to be in terms of battery power, be that in terms of grids or in terms of electricity generation, because the world's just going to need loads and loads more power.
00:13:44
Speaker
And we can get that more easily today using fossil fuels than renewable energy.
00:13:49
Speaker
But it's again, that sort of becomes another problem we're creating here.
00:13:52
Speaker
So we do need to see big, big breakthroughs in power supply to keep up with this great use of AI or else we're going to create some more problems in other areas.
00:14:00
Speaker
And to finish on, James,
Global Economic Variances in AI Impact
00:14:01
Speaker
one aspect that is less talked about at the moment is the distributional impact.
00:14:05
Speaker
Can you explain what that is?
00:14:07
Speaker
We've touched on it a little bit during this conversation, thinking a little bit about the winners and losers, but I think the big picture macro winners and losers are really important here.
00:14:15
Speaker
So if you think about the nature of jobs that are most likely to be affected by AI, it's generally lower skill, more repetitive services jobs.
00:14:25
Speaker
And generally, those are the jobs either taken on by people sort of early on in their career or the jobs that are offshore overseas.
00:14:34
Speaker
And this creates an enormous potential challenge if those sorts of roles are no longer either as plentiful or may not exist entirely.
00:14:41
Speaker
Now, starting with the sort of junior roles, there's going to have to be serious discussion and thought process about how you train people in this world.
00:14:48
Speaker
Even if I take my job, if I
00:14:49
Speaker
didn't need to do a lot of the grunt work in terms of collecting data or finding reports, all of those things, because AI was able to do it.
00:14:56
Speaker
Would I be able to build on that in terms of a skill set?
00:14:59
Speaker
And we use the example in the report of saying it's like trying to do an advanced mathematics degree without understanding the absolute basics.
00:15:07
Speaker
And that's a potential challenge.
00:15:09
Speaker
And that could create some impacts on the labour market, even in economies where there's some benefits.
00:15:14
Speaker
But the challenge then, I think, is more so in the emerging world where you've got a lot of economies whose either growth model today or potentially future growth model relies on exports of services, commercial services.
00:15:27
Speaker
And you look at examples like India and the Philippines, where that is currently about just under 10% of GDP and a huge amount of employment.
00:15:36
Speaker
And you think, well, actually, those are the jobs that are absolutely ripe for being removed by AI.
00:15:42
Speaker
And so not only could it impact those economies, it can impact other emerging markets who therefore don't have that as an opportunity to grow.
00:15:49
Speaker
And that could be problematic.
00:15:51
Speaker
And in the report, we look at some of the different metrics by which we can assess these risks, looking at the share of commercial services exports in GDP, looking at the nature of employment and how many jobs are in professional services and looking at AI readiness too.
00:16:05
Speaker
And a lot of emerging market economies
00:16:07
Speaker
don't have the right guardrails in place in terms of regulations.
00:16:11
Speaker
They don't have the tech adoption to benefit from the rollout of AI.
00:16:15
Speaker
Whereas in some developed economies, this is probably more likely to be a good news story where you've got the sort of tech capabilities, you've got a little bit more regulation in place and you're less exposed on the job front.
00:16:27
Speaker
I think actually the benefits could more so accrue to the developed world and hurt potentially parts of the emerging world.
00:16:33
Speaker
And that's
Conclusion and Subscription Reminder
00:16:34
Speaker
where you could get a really big global distributional impact
00:16:37
Speaker
and from AI, even if that global message is broadly positive.
00:16:40
Speaker
Well, James, there is, as you say in your report, a lot to weigh up.
00:16:45
Speaker
But thank you very much for answering my questions today.
00:16:47
Speaker
Thank you very much.
00:16:51
Speaker
That was James Pomeroy on the good, the bad, and the ugly side of artificial intelligence.
00:16:56
Speaker
James' report also features in the latest edition of Talking Points, our monthly wrap-up of the key reports related to the nine big themes we focus on here at Global Research.
00:17:07
Speaker
Other highlights this month include shifting demographics, the growing trade links between Asia and the Middle East, and the impact of U.S. tariffs on Chinese goods.
00:17:16
Speaker
If you'd like to know more, then please email askresearch at hsbc.com.
00:17:24
Speaker
So that's all we've got time for today.
00:17:26
Speaker
Thanks very much for listening to the Macrobrief.
00:17:28
Speaker
We'll be back again next week.
00:17:36
Speaker
Hello, Macrobrief listeners.
00:17:37
Speaker
I'm Fred Newman, and on this week's Under the Banyan Tree, we've got our eye on commodities.
00:17:41
Speaker
From climate change to geopolitics and the energy transition, there's a lot to pack in, all with a healthy dose of Asian economics.
00:17:49
Speaker
My guest is Paul Bloxham, our chief economist for global commodities here at HSBC.
00:17:54
Speaker
Listen, like, subscribe, and we'll see you Under the Banyan Tree.
00:18:17
Speaker
Thank you for joining us at HSBC Global Viewpoint.
00:18:20
Speaker
We hope you enjoyed the discussion.
00:18:22
Speaker
Make sure you're subscribed to stay up to date with new episodes.