Alexander Bick is a senior economic policy advisor at the Federal Reserve Bank of St. Louis, a macroeconomist with interests in labor economics, economic growth, and development. I visited there recently to hear Augustin Carstens, General Manager at the Bank for International Settlements (BIS), deliver the annual Homer Jones Memorial Lecture, introduced by the Bank’s President, Roberto Musalem.
Carstens delivered a comprehensive lecture on modern monetary policy frameworks. Alex in his discussion with me focused on a very different, and no less important topic: the rapid growth of Generative AI, and what how much it may boost labor productivity.
Alex and his co-authors used data from the first nationally representative U.S. Survey of Generative AI adoption at work and at home from the Real-Time Population survey (RPS). So dive in and hear, see what he has to say. And don’t worry about getting bogged down in the most technical aspects. He is making points that everyone can understand, and appreciate.
00:02:41:15 - 00:03:07:14
Home use trumps use rates at the office 00:03:07:16 - 00:03:28:18
I'd rather be asking how are you using it now and 40% of the people tell us they do <use it>… if you look at only at the workplace it's about 27%. And then at home it was I think 34%, 25%. So and then if, you know, some people use it at home, some only at work.
The first real tech adoption was was….the PC! 00:04:18:20 - 00:05:01:20
And if you think about what was the first mass market product for the computer, it was the IBM PC in 1981. And the first data that we had from the CPRS was from 1984, about three years later, after the first mass market product, kind of compared to what we have now for GPP and that adoption at work was very similar. Around 25, 26% versus adoption at home was much lower, which is maybe not surprising because PCs back then were very expensive and so very few people had that at home versus most of us have a phone and a phone in a computer. You have access to the Internet. So, you know, these all these GenAI tools, they're very user friendly, very easy to use.
AI is an easy product to try 00:05:02:01 - 00:05:21:14
People already are familiar with using Google like another search engine ‘Bing’ or whatever you name it. So in that sense, using chat GPT should be very easy. And that's why we think the usage rates have gone so much higher than before.
Census will mine more data on this 00:05:27:19 - 00:05:54:04
…Companies to say, well, we're using for this, we're using it for that. We're using for this project. We're going to make sure we give training, etc., that the companies are not moving as quickly on this as you might have thought. So actually there are data that the census is running a survey where they're trying to collect data on the adoption of generated by artificial intelligence at the firm level.
Deja vu all over again (for some) 00:05:54:06 - 00:06:37:27
And the latest data from early 2024 was only 5% of firms adopting it. Compare that to the other 26% to 27% of workers that you see that work as are doing that much more. And I mean, we were doing an interview with another outlet and they actually quoted the CEO of of Bucks, which is one of these data storage companies. And he was saying that that they saw exactly the same thing in the early 2010’s when they launched. He said people just adopted these tools at home. And you can see that in emails before firms had to kind of catchup. So that kind of can explain a little bit of that, you know, difference between what we see on the firm level <Vs at home>.
Corporate structure and control stifle innovation 00:06:37:22 - 00:07:22:09
Why is it is it because it's when you have certain jobs and certain procedures and certain ways you do things in a company that somehow introduce AI is like when you're home saying, oh, I'm going to see if I can find a how to get the cheapest flight and the best itinerary for to get to Mexico City or something like that. Whereas at work there's very specific things and there's hierarchies. And so you can't just dive in and say, Oh, oh, look at this, you know, Oh, it's look what I just did. What do you think, boss? You know? Yeah. I think part of it is some firms are concerned about, you know, data security, data privacy. So there might be reluctant to use it and workers might just use it for redrafting an email that is not security myself.
AI as editor and linguistic partner 00:07:47:00 - 00:08:08:13
And I think one of the biggest tasks is writing and communication. I think people like I, for example, like trusting in email, working on, you know, a summary of something, it is really helpful. Like I think like these two, that pretty amazing streamlining and getting rid of redundancy. And that's what we see. A lot of people are using it for that.
Where is all this heading? 00:08:32:19 - 00:08:51:15
So we actually have some concrete numbers on that. So we asked people on the frequency of usage. So we had kind of last week, how many days did you use it? And then on days that you're using it, for how long, use it to use it. So we ask about broad categories of combined. This information that we have on Facebook worked last week and I worked last week.
Who will be the users ? Who will it help? 00:12:40:17 - 00:13:13:21
So if you think about that might give you a glimpse into the future. And it's unclear yet, I think. But from these early comparisons, it looks about like the PC. And we know that benefited a lot. The high educated, high income earners. It increased productivity a lot, but it also increased inequality. But it also seems, though, in some cases that if you have you develop apps and you have a mobile phone, I mean, in poorer countries, in developing countries, that's a a source of technology, whatever you want to call that has been used, certainly.
But…there are low-income benefits as well 00:13:13:22 - 00:13:33:15
And you know, when people could track businesses, they could track different things. When they're starting out small business and not so wealthy. Place I agree. I think there's a lot of potential to help people in low income countries or lower educated people If they if they want to start a business, they might have not had the MBA training.
ALEXANDER BICK
Senior Economic Policy Advisor
Alexander Bick is a senior economic policy advisor in the Research division of the Federal Reserve Bank of St. Louis, which he joined in June 2022. He is a macroeconomist with interests in labor economics, economic growth, and development. His current research focuses on two broad topics: first, within- and cross-country differences in how much people work and how this decision is affected by public policies; second, trends in working from home in the U.S. and the consequences for job and geographic mobility. During the first two years of the pandemic, he ran a high-frequency online survey in collaboration with the Federal Reserve Bank of Dallas to provide real-time estimates of the state of the labor market.
Bick holds a master’s degree and doctorate in economics from Goethe University, Frankfurt, Germany. His research has been published in leading general interest journals in economics such as the American Economic Review, Review of Economic Studies, and Quarterly Journal of Economics. His research has been covered by national and international media outlets such as the Associated Press, Bloomberg, New York Times, Wall Street Journal, and Washington Post. After finishing his dissertation in 2009, Bick stayed for three years as an assistant professor at Goethe University. From 2012-2022, he was an assistant and associate professor of economics at Arizona State University, where he taught in the M.B.A. and Ph.D. programs. He spent his sabbatical in the fall of 2019 as a research economist at the Federal Reserve Bank of Minneapolis.
Share this post