AI and the global economy: unlocking key trends and opportunities

Jennifer Marin

Welcome to The Angle from T. Rowe Price, the podcast where curious investors can gain an information edge on today's evolving market themes. Sharper insights on the forces shaping financial markets begin here. In this season of The Angle, we're focusing on the exciting world of artificial intelligence, otherwise known as AI. I'm Jennifer Martin, a global equity portfolio specialist here at T. Rowe Price Associates.

And I'll be your host as we learn more about this intriguing and rapidly evolving area of the global economy and financial markets. In this series, we have explored the incredible pace at which it is being implemented and how several sectors are harnessing AI in their operations. In today's episode, we will take a step back and look at AI's possible impacts on the broader economy.

My guests today are Blerina Uruçi and Kimberly Johnson. Blerina is T. Rowe Price Associates' Chief U.S. Economist, and Kimberly is Chief Operating Officer of T. Rowe Price Group. Blerina and Kimberly bring their unique perspectives on the economy and technology to us today to share some of the trends that they are seeing due to AI. Blerina, Kimberly, welcome.

Blerina Uruçi

Thank you for having me today.

Kimberly Johnson

Jennifer, Good morning, and thank you.

Jennifer Marin

Before we get into the economic potential of AI, one of the first interactions individuals have had with AI has been through the use of Chat GPT in their personal lives.

How have you use Chat GPT in your lives?

Kimberly Johnson

Jennifer, thank you so much for that question. I think it's a really great way to get us started. I like to use Chat GPT as my writing assistant. I had to do a retirement speech, and I was just looking at a blank piece of paper completely stumped. So I asked for a retirement speech for a manager and it spit out a really good speech.

Now, it wasn't in my voice, so I had to do a revision, but the context and the structure was just really, really useful. And on a lighter note, I would say that I was curious coming into this

podcast, so I asked Chat GPT to write me some haiku. I like the poetry that it comes up with, and I asked it to write me a haiku on podcasts, and it was pretty good.

And then I asked it to write me a haiku on AI, and it was even funnier. And then I asked it to write me a haiku on podcasts about AI. And here's what it came up with.

Bytes of knowledge flow voices weave tales of AI minds and algorithms grow.

So there you have it.

Jennifer Marin

Beautiful. I love it! Blerina, do you have anything to add to that?

Blerina Uruçi

Something much more mundane than Kimberly. I like to use it to help me save time, multitasking in my personal life. So recently I took a family road trip, and my AI model was my travel agent of sorts. We had three generations traveling with us, so as you can imagine, we had a lot of boxes to be checked, like large enough accommodation for all of us, something relaxing and kid friendly, mountain biking paths, rock climbing options.

And we had to plan this trip while doing our full‐time jobs and looking after three young kids. So naturally I turned to my friendly AI model. After a few iterations, I got an itinerary, accommodation options, attractions to visit in each area. And I have to say, having done the trip, the feedback was good enough, considering some of the travelers in my family are quite hard to please.

But unfortunately, we're not there yet where I could give it my credit card details and ask it to make reservations. But I wonder how far that is in the future.

Jennifer Marin

Yeah, probably not as far as we think. And you know, it's a great segue into how, as individuals, we've grown more comfortable with AI technology. We are beginning to see more widespread adoption of AI technologies in business applications. So could you discuss the potential for AI to drive productivity gains, innovation, and economic expansion?

Blerina Uruçi

This is a very good and very broad question, and I think the key to answering it is going to be how quickly and how widespread the adoption is going to be. So AI, as a technology itself has been in the making for many decades. But the release of Chat GPT in late 2022 really brought this technology directly to the consumers and really made AI part of dinner table conversations and C‐suite meetings in equal measure.

Then you look at Google search interest in AI. That really also took off in late 2022. This is when the hype started. And I have to say that hype hasn't really died down yet. And I I think there is something that makes AI different from other innovations that we have experienced in the past is the fact that it's easy to access from our phones, it's publicly available; it's free; and I think this is going to make its diffusion faster and deeper than what we've seen in the past.

So when we look back in history, there has been this predictable pattern of innovation shocks. The sequence has been that you give it to developers. They have access first, then businesses develop machines, and they develop applications for workers to use it. Then the technology goes to the worker, right? So the sequence is developers first, then businesses, then workers.

What we're seeing today is that AI has almost skipped a step. It has gone directly from developers to consumers to the workers. And we've already discussed examples of how we're using it in our personal life to plan holidays, to write speeches. So because of this, I think businesses have to move faster for two reasons. They want to understand the technology in order to to protect data and proprietary business intelligence.

They also have to speed it up by building on publicly available tools for the specific needs of the business, of the workers to enhance their productivity. So these are two very strong incentives for businesses to adopt technology and to adopt AI much faster than we've done in the past.

Now, having said all of this, this is a very useful framework for me as an economist to think about technological innovation broadly and about AI more specifically.

But Kimberly leads the operations side of our business with thousands of employees. What are you seeing in your day‐to‐day job Kimberly?

Kimberly Johnson

Yes, Blerina thank you for that. It's true. I I love what you had said about having AI at people's fingertips because it's free, it's available, it's in their phones, it's mobile. We definitely have seen that as a leading edge. You always get early adopters in your workforce.

But from a business perspective, there's really three different, really important interactions. It's the way that you engage with your customers, the way you engage with your daily tasks, and the way you generate value. And so I think getting your early adopters to get comfortable with your tools and to be the folks who are really the trailblazers running ahead is is very helpful.

But at the same time, from a corporate perspective, you do have to put the guardrails in because your risk appetite at home is really different than your risk appetite at work. So the kinds of things that we're doing is running sort of controlled pilots where we can observe the data and the interactions. We have those early trailblazers as our best users and that helps us find exactly where we need to augment the guardrails and the controls in a managed and well responsible way.

Blerina Uruçi

So I could also share how we use AI on the research side of the business. Internally, we've launched our investor copilot as a tool to enhance the investment process.

The investor copilot is bringing a very, very long history of research at our fingertips, and it is connecting the insights between the equities and credit angles really seamlessly. By doing this, it's creating space for our analysts to do more deep thinking. So every week that passes in our copilot, we're figuring out ways in which AI can be harnessed to connect our investment professionals to our firm's wealth of knowledge that we've built historically.

Kimberly Johnson

That's right. Taking our proprietary data and using that to augment all the open‐source tools and data that is available. It's where we can fine tune our models and make sure that we use script retrieval, augmented generation that doesn't just spit out answers or haiku or travel plans, but actually comes up with real research‐driven conclusions that are attributed.

We can go through and click through to the source using some of those tools, and that means we can validate and verify that the information is credible, really important in a business context.

Jennifer Marin

Well, I know for years we've been saying data is a strategic asset, and I know every organization overspends on cybersecurity, and now we're probably overspending on AI, but in some ways it's energizing to hear the AI, the AI spend because there is some phenomenal insights that this technology is going to allow us to harness

So how are you measuring productivity gains?

Kimberly Johnson

Oh, that's a great question, because productivity is one of the first things people look for when they start implementing this in a business setting. You look in two areas really: In the text arena, there's kind of obvious everybody's using chat bots and assistants and that to help in response to inquiries. You also have content writing.

You can use it for search. You can use it for analysis and synthesis. Lots of ways to get productivity. The code arena is one of the most measurable places that we do it though. You get uplift in terms of code generation. Generating new datasets to test the code creation of documentation is huge. Our developers spends so much time documenting and so much of that is rote and repetitive that they've found that with the AI tools that are available, we can get 30 to 40% gains in the developer productivity when they're coding.

Jennifer Marin

And I know in speaking with companies, many are deploying this software as well, and they're seeing some of their developers get nearly 80%, like it's making their best developers, even better‐‐super developers.

Kimberly Johnson

That's right. And the super developers are definitely emerging. It's almost a barbell where you've got your super developers who get a lot more productivity because they're excellent at adopting the tools. But for your lower‐performing coders, it's also really helpful because it takes a lot of the time out of the things that they find to be repetitive and difficult to do.

Jennifer Marin

Yeah, and hopefully makes them happier. I think that's what we're going to end on. Hopefully at some point.

Blerina Uruçi

Productive workers are happy workers.

Jennifer Marin

Yes! I love it. Alright, well, that all sounds very promising, but what are the bottlenecks that economic theory predicts that could hinder the growth‐enhancing effects of AI?

Blerina Uruçi

So I can think of broadly three areas here. For example, adapting AI and building the tools that can meet the specific needs of the businesses can take time. Sometimes this may not be as fast as we want to. Many business leaders and teams are still figuring out ways in which the power of AI can be used to increase worker productivity.

Some of the tools that will really help to achieve that goal are not built yet. The second area is that we've spoken a lot about enthusiastic early adopters so far, but inevitably there are always workers who resist innovation, especially if it's seen as a threat or if it's not introduced in a user‐ friendly way. Also, if education and training of these workers is not moving as fast as the development of AI is happening, this can slow down adoption and the growth‐enhancing effects of AI.

And then the final one I want to close with is that there is another scenario in which society may actually choose to create some bottlenecks. For example, through regulation, because we still do not understand the ways in which AI could be harmful. This fear of the unknown or the potential for misuse of AI could lead to less experimentation.

Although I think this will in the end likely be offset by the business's drive for innovation and to remain competitive in this field. So I've given these ideas. We've remained very much in the theoretical framework, but I think this is a question where economic theory can be really informed by executives, so in areas where the data that is being handled is sensitive.

So, Kimberly, from your experience so far, what have been some of the more challenging workstreams for AI adoption for you?

Kimberly Johnson

Thank you, Blerina. And you're right, I think the concept of almost self‐inflicted bottlenecks is an important one for us to touch on in a minute because we like to think of using almost like a traffic pattern for this one. Some are green. Go, go, go. And I think from an enterprise perspective, the way to really get your broader workforce in the go, go, go mode is around adopting it through tools that they already have for productivity, and the embedded AI components within those tools, they’re already familiar with is a great way to get people to use things in a controlled manner that you can feel really safe and secure about. I think in the yellow space, that's where most of what we do comes into play though, right? There's the augmented research that we're talking about.

There's some of the coding that we do. With each of those types of engagements, you need to check and make sure you've got the right guardrails, the right data management, the right controls. You've got to understand the data governance, the cyber security. You've got to make sure the infrastructure is scalable and can handle the kinds of computation that are going through.

So there's a lot of places where you have to do check ins, and I think that's a little bit of a bottleneck. It's important for companies to build out their data governance and their infrastructure in concert or maybe even a little bit ahead of their AI ambitions to make sure that it can keep up. And I think that is a really essential component of going hand in hand.

The enablement along with that innovation and experimentation.

Jennifer Marin

That’s really helpful. It is fascinating to understand all the work around what generative AI will do and in some cases what you both just said. We're pivoting now to what we want the technology to do, and that is, you know, I think the future of what this technology can do from a productivity standpoint. So if we move on to our next question, some people are concerned about how AI is redefining the complementary relationship between technology and human labor across various industries.

Do you see scenarios under which AI could either complement or substitute traditional roles?

Kimberly Johnson

So what we're doing here at T. Rowe Price is very much human in the loop, right? It's a complement. We’re augmenting people's capabilities. We’re not really substituting for judgment. I think there's a lot of things that we can do to substitute. It’s the lower level work the kinds of things that are just highly redundant, very rote, menial tasks like reconciling data sets that maybe you could get a machine to substitute doing that for you.

I do think that there's very high value work that you want your humans doing in the first place, and the opportunity is to make sure that you give them the bandwidth to do that high value

work by making it much easier, simpler, and faster for them to do the things that aren’t truly value add.

Blerina Uruçi

So what I would add to Kimberly is that when we look at this from an economic theory perspective, we can have two extreme views of how innovation, and in this case AI, is going to affect the labor market. You can have dystopia on one hand, so the majority of the jobs will be destroyed, and then you can have utopia on the other hand, which is we will all be better off, we'll be more productive, we'll be happier in the workplace. And then the question is, what do American workers think about it right now? And a recent Gallup poll suggests that three quarters of Americans are closer to the dystopian view of the world. People are actually concerned about this new technology.

And this is where leaders such as Kimberly come in to encourage people to adopt it, to make it part of our day‐to‐day work. And then looking back at history, what does history tell us about innovation? Actually, the outcomes should be closer to utopia. That means that technology does not reduce net jobs in the economy, does actually redistribute economic advantages and skills.

But, and there will be winners and losers, and this is how we need to manage things carefully. And then the historic example that I would bring into the conversation is that at the onset of the Industrial Revolution, more than 90% of the world population was working in farming. Back then, we could not imagine a future where we're all sitting in offices, working on our computers, recording podcasts, and adding value to the world economy in such different ways.

So right now, the future ahead of us is probably beyond our imagination, and we should all start reading a bit more sci fi for inspiration here.

Kimberly Johnson

I think that's great with the sci fi, and oh my gosh, I couldn't agree more. This goes back to that conversation we were just having about happy workers. We do measure satisfaction along with productivity, and what we've found is that people who've adopted the AI tools have a much greater job satisfaction. When it comes to convincing people, I think the easiest way to convince is just to reframe in terms of there's no reason to jump to the dystopia where, you know, machines are coming to take our jobs.

If you ask someone to please walk you through the worst part of what they have to do as part of their job responsibilities, they will usually sit down and tell you and they have a pretty good list. And if you ask, well, what can I do to help automate some of that? Is there anything that I can assist you with in terms of eliminating some of that work?

And if you bring the right tools, you bring the right people to help with process reimagination and implementing some of these really important tools that are available, you can quickly start

decomposing the worst part of people's jobs. And this turns into something that they're actually looking forward to.

Jennifer Marin

Yeah, it sounds like a lot of people will be open to change as we redistribute economic advantages and scales.

Kimberly Johnson

Yeah, basic examples, right? Like librarians used to work in the Dewey Decimal System with card catalogs and only with paper books. Librarians now have all kinds of information at their fingertips, and they've broadened their skill set and what they're exposed to to help bring that to other people, too.

Jennifer Marin

Maybe we should have asked a librarian to join us to give us their sci fi recommendations, because my mind immediately went to Margaret Atwood, who I love. But there are so many other good ones. We could spend a whole time on that.

Blerina Uruçi

And and really jokes aside, I think the skills of the people already in the workforce and the future generation of workers really needs to include how to engage with AI models to enhance our productivity. I think this is important. And then we have many academic papers that have looked at industry level data to come up with professions that will be most affected by AI so we can prepare our education system in how to train these workers.

The UK government has also come out with a report and what this research has in common is that it's listing the professions that are in the business sector, in the services sector, such as consultants, lawyers, investment analysts such as myself, education professionals, writers. These are all jobs that will be highly impacted by AI. Now, like Kimberly, I don't want to take the pessimistic view here.

My more optimistic interpretation is that these jobs will become more interesting, and they'll become more fulfilling, we’ll do the things that we enjoy most about our jobs and we should get AI to help us improve on this efficiency and productivity. So right now, I cannot imagine a world where we will not need lawyers or writers, but I can see a future where those professions will become more productive.

And maybe we need fewer workers.

Kimberly Johnson

That's right. I think the skills that people develop for those new jobs is really important. I saw a report out from LinkedIn just a few weeks ago and it listed the top ten skills in terms of demand for 2024. Number one, of course, was artificial intelligence and machine learning. It talked about engineering, design, implementation, and optimizing systems. Number two, was

cybersecurity. Again‐‐ guardrails, we've been talking about making sure you keep your data and your infrastructure safe.

Number three was data science, really understanding how to analyze the information and build the models. And number four was cloud computing, and that's the essential infrastructure you need for AI. There's so much synergy in those four skill sets and it tells you so much about where we're headed. I think the um, the other thing to think about is how fast this is all evolving.

I think prompt engineering is a new skill set that I see everywhere now, and that didn't even exist a few years ago. That that shift has been happening over the last decade. Half of the Ivy League schools have computer science as their most popular major, and CS degrees at some of these schools have grown three‐fold over the last ten years.

That's a remarkable demand for computing and technology skills. So I think the world is evolving and people are working their fastest to keep up. I'm excited from the enterprise level about what we can do to help people to build some of these new skills for the future.

Jennifer Marin

What strategies can organizations and workers adopt to navigate the transition towards a more AI‐integrated economy?

How can workers and companies prepare for the shift in skills demand that AI is expected to bring about? And I know, Kimberly, you started a little bit on this, but maybe continue on that topic.

Kimberly Johnson

Thank you, Jennifer. I think it really raises the question of how quickly are we able to teach people new things and teach them a way they can apply them. And not just for their entertainment, but really for driving that business value. And for me, it's a question of how do we create the parallel tracks for development, growth, and learning that go along with these really fast innovations.

We found that you can do labs and sandboxes. You've got to have places for people to put their hands on and do real work using the real tools for them to truly learn it. Nobody learns unless they get hundreds of repetitions is is the is the natural science. Right? So once you get over and over and over again an opportunity to get not just familiar, but to really become facile in working in these things, you can you can start to get adoption and usage and implementation into the day‐to‐day work.

So I think getting the right training, the right development, all of those things are great. From a classroom perspective, the exposure is excellent, but until you match people and work and skill sets together in a way that they can come together and develop real, valuable, productive pieces of their business environment, it's hard to say that they've learned it.

Blerina Uruçi

Another important aspect to discuss here is like, how are skills affected in the workforce?

And this is going to depend on adoption and if it's going to be a sequencing that is, if we think about adoption and how it affects the skill set of workers as a S‐shaped curve, early on, we have worker replacement. This is where technology can replace or can do those jobs more effectively, more cheaply. Then we have worker enhancement.

This is the bulk of what has happened through historical innovation and then the last stage can be job creation. This is where the new jobs that didn't exist before in the economy come in.

What is happening with AI is that this traditional sequencing might not be working at all, and everything is being bunched up together. A good example here would be health care.

AI can really enhance the work of health care professionals. It can diagnose tumors with complex imaging in much less time than a human. But you still need a human doctor to make the final decision to make a differentiated diagnosis to speak to the patient. So what happens here is AI is actually enhancing and saving time for doctors by making the diagnosis faster.

At the same time, the use of AI will create demand too for teams within the health care system that can train and further improve efficiencies in these AI models. And so here comes prompt engineering that we've already discussed. So I think with AI, we're going to experience everything at a much, much faster time schedule.

Kimberly Johnson

That's right. You wouldn't think you would see a big demand for engineers and software developers in the health care space, for example. But you give a fantastic example of how that really can be leveraged in almost any industry.

Jennifer Marin

There's a lot of excitement about the economic opportunities presented by AI, as well as concerns about its misuse. How can we balance this enthusiasm with the challenges of ethical oversight?

Kimberly Johnson

Well, we're always cognizant of the ethical limitations. Right? You've got to ensure that you have very good visibility. It goes back to traceability. It also goes back to the data sets and the data management, making sure that you have reliable data that is sustainable over time and free from bias. From a corporate perspective, I think it's really important to make sure that what we're looking at is not just what people put into the AI prompts, but what comes out of them.

How do you have a way to validate some of these answers? How do you take care of you people? How do you prepare them for the disruptions from the technology that we're

expecting? Just as Blerina was explaining, there's all kinds of opportunities to augment our ability to do our jobs well.

Jennifer Marin

But where my mind went when you were talking about that was in some ways it has like a Reagan‐Gorbachev kind of vibe, which is trust but verify.

Kimberly Johnson

Yes, it's a very trust, but verify.

Jennifer Marin

And so it is still early days and that's why the source attribution is so useful, because then you can go back to see, you know, how real is this?

Kimberly Johnson

That's right. And you measure the number of hallucinations that come out of our models. The answers that look right but actually are not accurate. And so part of this testing is making sure that you can diminish the hallucinations to the point where you can rely on the model with a level of confidence, but also identify what those hallucinations look like through that verification of the source data and what's behind the answers that it spits out.

Jennifer Marin

And maybe just for our listeners’ benefit, can you maybe just share, generally speaking, what a hallucination is?

Kimberly Johnson

Yes. The AI tooling generally will give you great answers depending on the data that it has. But if there's data missing or something that's inaccurate, it can take that data and analyze and synthesize it and give you an answer that seems like it is plausible but is not actually accurate.

Blerina Uruçi

And that's why you need matter experts to validate the answers of AI. And that's why when we talk about how it's making us more productive, we're not at the stage where you can copy and paste everything that an AI model is giving us. It’s we use our own critical thinking. Because when you look at how AI operates at this moment in time, critical thinking, it's one of its disadvantages.

It's what's its weakest qualities.

Jennifer Marin

Blerina, you're going to make every teacher in America so happy that they can still teach critical thinking. You can't. You can't learn just by Chat GPT. You have to think it through. You probably you’re going to get an award from someone soon.

Kimberly Johnson

And AI will not let you off the hook.

Jennifer Marin

Yeah, right. So maybe pivoting to another area, Blerina. In what ways could AI driven innovations disrupt traditional economic models and forecast that many decisions rely upon?

Blerina Uruçi

This one is very close to home, right? This is my bread and butter. So forecasting is hard. This is both a very broad statement and it's a broadly correct statement. We get things wrong all the time and then within that universe, one of the hardest things to forecast and predict is a productivity shock. And so what we've experienced since the global financial crisis is a period of very slow productivity growth, productivity underperformance.

And then you couple that with a demographic drag from an aging population. This has really slowed potential growth for the U.S. economy and the world economy more broadly. So then the big question is we're not expecting potential growth to really accelerate in the very near future, but could AI provide the next productivity shock that can make us all wrong because we can have our cake and eat it. We can grow faster, we can remain at full employment and we can keep inflation under control.

And the answer is maybe yes and maybe sooner than many of us expect.

Kimberly Johnson

The rate and the sustainability of those productivity shocks. It's hard to imagine and definitely hard to predict. In financial services, we generate so many forecasts and models and different ways to think about the future. I wonder if one thing the generative AI might be able to do is to help us in being more creative, more extensive around the scenarios that might just be possible.

And I think about stress testing and forecasting. We generally anchor on the things we know. We think about the scenarios that we've lived through and whether or not they're likely to happen again. We think about previous crises when we build all of our new stress. And so the one thing I the really interesting question to ask, if you ask an AI model to come up with something that doesn't have bias around all of the historical things that generate our forecasting ideas, is there a possibility that we could come up with new risk scenarios?

Blerina Uruçi

I mean, I love this question, Kimberly, because this really brings us to the point where we can make the distinction between what is possible now and what could happen in the future. So right now, I feel like we are at the stage of development where AI's learning from the collective human history and producing answers based on our collective knowledge.

This step towards creativity and critical thinking that we discussed is not something that current models are very good at. Today we're talking about critical thinking in the context of financial

models and stress testing scenarios. If you have this conversation with writers, they could come up with a similar argument about how creating unique artistic forms of expression are hard for AI.

If you speak to comedians, they'll tell you that AI can't write funny jokes the way that humans do. So this is where we are now. But that doesn't mean that AI is not going to develop critical thinking that expands beyond what the history that it is studying suggests. So that could be the future, that we could ask AI to stress test scenarios beyond our imagination.

Jennifer Marin

That leaves us with a lot to think about. Blerina and Kimberly, thank you very much for today's fascinating discussion and for helping us think about some broad implications of AI implementation. I've really enjoyed it, and I hope our listeners have too.

Blerina Uruçi

What a great conversation today. Thank you both.

Kimberly Johnson

It was terrific to be here.

Jennifer Marin

We will continue having conversations about the importance of AI in our next episodes. We hope you can join us. For me, the key takeaways from today's discussion are broader understanding of the wider economic implications of AI on growth, employment, and productivity; the importance of policymakers balancing the economic opportunities presented by AI with the challenges of ethical oversight; and the potential of AI‐driven innovations to disrupt traditional economic models and forecasts.

And thank you for listening to the angle. We look forward to your company on future episodes. You can find more information on artificial intelligence on our website. Please rate us and subscribe wherever you get your podcasts.

Disclaimers

This podcast episode was recorded in April 2024.

This podcast is for general information and educational purposes only, and outside the United States is intended for investment professional use only.

It does not constitute a distribution, offer, invitation, recommendation, or solicitation to sell or buy any securities in any jurisdiction, or to conduct any particular investment activity.

This podcast does not provide investment advice or recommendations, nor is it intended to serve as the primary basis for an investment decision. Prospective investors are recommended to seek independent legal, financial, and tax advice before making any investment decision.

The views contained herein are those of the speakers as of the date of the recording and are subject to change without notice. These views may differ from those of other T. Rowe Price companies and/or associates. Information is based upon sources we consider to be reliable; we do not, however, guarantee accuracy.

Publications referenced in this podcast include:

“Business in Society Report,” Bentley‐Gallup, 2023.

“Top 10 In‐Demand Tech Skills for 2024,” Fuel Recruitment, accessed via LinkedIn, 2024.

T. Rowe Price has been testing the use of AI models across a variety of job functions. Activities mentioned in this podcast have been in beta testing as of April 2024. T. Rowe Price may alter, expand, or discontinue its testing at any time.

This podcast is copyright 2024 by T. Rowe Price.