Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=

# The Microstructure of AI Diffusion: Evidence from Firms, Business Functions, and Worker Tasks

Authors: Kathryn Bonney, Cory L. Breaux, Emin Dinlersoz, Lucia S. Foster, John C. Haltiwanger, Aditya A. Pande  
Discussant: Martin Beraja  
Video: https://www.youtube.com/watch?v=VdvT0JzwMHU&t=21763s


## Talk (06:02:43 – 06:23:58)

[06:02:47] structure of AI diffusion. >> Just making sure this works.  
[06:02:54] Just this. [clears throat] >> Okay, great. Thank you. Thank you very much. Um this paper is very much a first look um at the uh micro data from about  
[06:03:07] 120,000 firms collected in the second supplement of the uh business trends and outlook survey. Um the data are um weighted using firm and employment rates  
[06:03:20] to ensure that the estimates are representative of the US uh business employer business population.  
[06:03:28] uh I should say that the paper was written in a very short amount of time to accompany uh the public data release on April the 23rd and we would like to build on the analysis here and add more  
[06:03:40] uh uh elements to the paper. uh but for today I'm just going to go through some of the very basic descriptive patterns emerging from the data so far and these results will pertain to general diffusion patterns both time series and  
[06:03:54] cross-sectional um firm uh firm level versus f worker level use of AI uh both assess by firm respondents and I should uh make a caveat here that uh none of these  
[06:04:09] estimates we're going to talk about on the worker side will be at the worker level. Uh um BTOS is not a worker level survey only a firm level survey. Uh so the best way to see these these results  
[06:04:22] is a firm level assessment of both organizational adoption and worker use patterns by uh by the respondents at the firm level. Um the ideal survey should of course strive  
[06:04:34] to collect data uh matched at the firm and worker level but we didn't have that luxury. So we're going to look at also diffusion within firm AI deployment across business functions and worker tasks impact within firm task effects  
[06:04:48] employment and capital substitution and finally we'll relate in a non-causal fashion um AI use aspects to uh to firm outcomes how the breadth of functional use worker task use and complimentary investments relate to overall firm  
[06:05:03] performance and changes in sales and employment.  
[06:05:08] So uh since November 2025, PTOs has fielded revised core and supplement questions on AI. Before the revision uh firm level AI use was elicited using these two questions, these two core  
[06:05:21] questions. One was about the current use in the last two weeks this business use artificial intelligence in producing goods or services and accompanied by a forward-looking question uh expectations during the next six months. Do you think  
[06:05:35] this business will be using AI in producing goods or services? Examples were given uh accompanied by a definition of AI and further examples.  
[06:05:46] Um [snorts] the wording focused on AI use in in producing goods or services.  
[06:05:51] This was intentional. This was meant to be an umbrella term, a deliberate filter to capture significant AI integration into main business activities related to uh provision of goods and services. uh  
[06:06:04] and tried to delineate this important activity from auxiliary incidental or peripheral users with little measurable impact. But of course, geni revolution changed uh some of this view. Um  
[06:06:18] functional diffusion especially um in functions like marketing, IT and finance enabled by Gen AI challenge the uh legacy terminology. um these functions  
[06:06:32] uh may not be necessarily seen as being in the realm of producing goods or services by at least some respondents and we obtained some additional feedback uh from the last FC meeting in 2024.  
[06:06:46] uh and further cognitive testing in 2025 revealed that respondents in firms focusing on activities like intermediation, marketing or administrative uh functions struggled with the term uh producing goods or  
[06:06:59] services. As a result, we revised the questions to have a better assessment of how firms formally integrate AI into their business operations. So we simply replaced in producing goods or services  
[06:07:14] with any of its bisonous functions in both of the questions keeping everything else intact and this this has had an effect on our estimates as we'll see shortly. Um and the second BTOS AI supplement goes  
[06:07:28] further and adds new questions uh on firm AI use in 15 business functions, worker AI use in workrelated tasks in support of business functions. Uh more general worker generative AI use in  
[06:07:41] workrelated tasks. Uh geni use in nine different tasks and uh types of uh worker task effects augmentation, substitution versus creation of tasks.  
[06:07:55] uh it also retains questions on the impact of I of AI on employment effect uh AI capital substitution intensity of AIdriven text replacement and and other aspects. So overall both the core  
[06:08:10] questions and the new supplement capture several extensive and intensive margins of uh AI use and impact. So here's a quick look at the time path of AI diffusion based on the core AI  
[06:08:22] questions. um firm weighted use rate rises from about 3.5% to 10 to 11% before the core uh questions were revised and then jumps to 18% post  
[06:08:36] revision. Uh expected to reach about 22% in the next 6 months. Uh employment weighted rates sit around 35% and expected to reach 47% uh in the next six  
[06:08:49] months. uh you can see the jump is sizable in relative terms but not so much in absolute terms. Uh it didn't take us to 50% or 80% adoption rates. Uh and you can see the gap there in the  
[06:09:03] graph towards the end and that corresponds to the laps in uh government funding after which we immediately resumed releasing u estimates.  
[06:09:15] Okay. So uh next is AI used by sector. uh AI is diffusing very unevenly uh across sectors and AI uses concentrated in uh knowledge and human capital intensive sectors such as information  
[06:09:30] professional services finance and accounting they have about 30 to 40% adoption rate firm weighted nearly double the national average and even more pronounced if you look at the employment weighted results in the uh  
[06:09:44] graph below uh the gap is uh even more clear there. Uh on the other hand, physical output and trade sectors, manufacturing, retail and so on so forth tend to lag. Uh  
[06:09:57] in terms of the near future, high use sectors are expected to exceed about uh 50% use rate in employment weighted terms. At the same time, some leaguered sectors uh such as retail, manufacturing, wholesale show near-term  
[06:10:12] high uh high expected growth rate in employment weighted use rates.  
[06:10:22] Um AI use by firm size is similarly concentrated. There's a firm firm size AI use gradient as you can see in both of the graphs firm weighted and employment weighted. Uh use rates rise  
[06:10:36] from about uh about 18% to 13 31% going from small to large sizes. uh both uh you know and also the rise is evident in the employment weighted uh uh rates as  
[06:10:49] well. Um but the important thing to recognize here is that we get even higher use rates in the right tail use rates in 1,00 plus employee firms reach about 35% and on top of that there's an  
[06:11:04] interesting size sector interaction uh aspect of of this. If you look at really large firms in AI heavy sectors, we reach rates like 57%  
[06:11:17] in current in current use and 68% in expected future use firm weighted and employment weighted numbers are even larger. So definitely an important aspect of the diffusion of AI is this  
[06:11:29] this interaction between uh AI heavy sectors and and large large firm sizes.  
[06:11:38] And I should note that use rates for the next six months are expected to be higher across all sizes especially for uh large sizes. This indicates that the uh adoption discrepancy between small  
[06:11:50] and uh large sizes large firms are uh is increasing but we should do more to quantify this increasing inequality.  
[06:12:03] So the next is a quick look at the AI used by firms versus workers. Again, as I mentioned, this is at assessed at the firm level. Uh so here's a generic representation of  
[06:12:17] uh AI use by by firm a and and workers. uh a firm could have firm level systemic AI as background automation and it could provide output to uh someone like worker  
[06:12:30] one who does manual task. In this case firm is using AI but the worker is not necessarily using AI or interacting with AI in an organic fashion. Uh in another example um the interaction could be both  
[06:12:44] ways right the AI supplies some information and worker uh gives feedback in that case the worker is is AI augmented there could also be shadow AI use within the firm uh where a worker  
[06:12:57] uses personal personal AI tool and also there could be workers using just simple uh traditional work so what we find is that 18% of firms use AI I in the last  
[06:13:10] two weeks and in 23% of firms workers use AI in work rellated tasks. So the two rates usually overlap to a large extent. Uh however what we find is that  
[06:13:22] conditional on firm use 19% report no firm no formal worker uses. This is a interesting case uh but could be possible when you consider um AI automation in cases where AI used  
[06:13:36] in for example fraud detection, credit screening, predictive maintenance, it may just generate some output that is passively used by workers uh who are basically interpreters or exception  
[06:13:49] handlers. There could also be a vis visibility bias issue here. uh some of the respondents at the firm level may not be aware of the worker level uh use aspects.  
[06:14:01] On the other hand, condition on worker use a larger fraction of firms report no formal firm use. So the basic conclusion from these uh descriptive patterns is that AI seems to be diffusing via dual  
[06:14:15] pathways. Both top down and bottom up adoption patterns appear relevant.  
[06:14:20] But obviously we have no data on the dynamics of adoption and and the mechanisms directly. Uh no questions on those quick comparison with other estimates in the literature. Uh others find much  
[06:14:34] higher rates like 70% firm level uh firm weighted 80% employment weighted and 40 to 45% at the worker level.  
[06:14:43] uh some survey samples are more representative of large firms, AI heavy sectors and specific respondent types such as executives.  
[06:14:52] Question wording, AI deficient examples and so on so forth also also matter. uh one important point here is that concentration of AI used in large firms and certain sectors imply caution in  
[06:15:05] comparisons when we restrict attention to subsamples of BTOS data similar to the samples used in other surveys we find much higher rates for example large firms in AI heavy sectors reach 50 to  
[06:15:19] 60% adoption rates 60 to 70% employment rated on top of that if we restrict further to top executive and owners we get 70 to 80% uh adoption rates and uh  
[06:15:33] even though our survey is not worker level when we uh employment rate the worker adoption rates uh worker use rates we get numbers like 40% which are not too far from the estimates out  
[06:15:46] there. So the main uh conclusion here is that various surveys and estimates are not necessarily inconsistent but but complementaryary uh in terms of the business functions  
[06:15:59] very quickly conditional uh firm using any AI in any business function sales and marketing lead at 52% strategy and business development follow uh and that  
[06:16:12] is followed by IT uh AI using it at the bottom are distribution production, sourcing and supply functions. This is partly driven of course by the specialization of business functions in  
[06:16:25] industries or firms. Uh and we see that growth is expected in all functional use rates uh in in the near future. uh one other interesting observation is that  
[06:16:38] functional use seems to be narrow and concentrated in general in 57% of firms with functional AI use uses at most in one to three functions  
[06:16:51] similar concentration in uh genai use in worker tasks writing and editing uh has the highest rate 85% information search uh and technical help come second followed by document ment  
[06:17:05] analysis uh coding, debugging, customer support, tutoring and learning uh are at the bottom. Again, partly driven by industry and business niche. Uh for example, writing and editing is common  
[06:17:18] across um uh many firms, but coding can be more specialized to tech firms and other other firms that engage in uh software.  
[06:17:30] Again, just like as in the case of business functions, we find that task use is narrow and concentrated. In 65% of worker genai use cases, use is limited to at most one to one to three  
[06:17:42] tasks. We also look at operational investments to implement, enable or use AI. Uh and we find that most firms actually report  
[06:17:53] no adjustments or investments about 64%. Uh as you can see in this uh response category at the top, employment weighted rates are far below the uh firm weighted rate suggesting that uh these could be  
[06:18:08] relatively small firms using off-the-shelf uh AI. Uh most common adjustments uh develop new workflows, train current staff, purchase cloud and uh cloud storage and computing power. uh  
[06:18:22] and these seem to be prevalent uh in the relatively large firms uh based on the uh graph again where employment weighted rates exceed the firm weighted rates for for these uh for these specific response  
[06:18:35] categories. So um uh these adjustments and operational investments may signify more formal use, structural integration, more pronounced  
[06:18:49] performance and employment effects and initial implementation frictions and gradual accumulation of AI related intangible capital with implications for the productivity uh J curve. I'm going  
[06:19:01] to skip this part for uh for the uh in the interest of time. So I'm going to talk about effects on tasks uh capital uh and employment. Uh so the firms were  
[06:19:13] asked to uh uh describe the effect of AI use on worker tasks in terms of augmentation, substitution and creation of new tasks and firms had the option to select all that apply. uh using these  
[06:19:28] responses we constructed mutually exclusive task effect combinations conditional on some effect and the striking picture uh here indicates that augmentation dominates 66% of firms  
[06:19:42] using AI uh you use it solely for augmentation uh purposes uh and task replacement intensity seems limited condition on replacing tasks 71% of  
[06:19:55] firms replace small number of tasks and 7% replace a large number and uh overall we see low incidence of employment change only about 4.3% of AI using firms report any AIdriven change  
[06:20:09] either increase or decrease uh and 2% report only 2% report employment degrees net employment degrees on the other hand capital substitution seems to be more prevalent 16% of AI using firms uh  
[06:20:24] report substituting software and equipment with AI. So it seems like more than the employment effects, AI is having uh some effect, some tangible effect on uh capital substitution and new capital  
[06:20:38] installment. The final analysis that I'm going to talk about is a simple regression setup to relate AI use to firm outcomes. For this purpose, we created three diffusion indices capturing different dimensions  
[06:20:53] of AI use. One is a functional use breath index index which is simply the number of business functions using AI divided by the number of available functions. And we did a similar construction for worker task task use  
[06:21:08] breath and operational investment breath. And then we relate these three different aspects of AI use or operational investments to enable AI use uh jointly to qualitative BTOS firm  
[06:21:22] outcomes in a linear probability model framework with rich fixed effects controlling for size class three-digit next industry state and survey period.  
[06:21:33] The outcomes are constructed using the data already collected within BTOS above average overall performance dummy, sales increase dummy, employment decrease dummy and then AIdriven employment  
[06:21:45] decrease dummy. And the summary of the results are shown here on the performance and sales side. Uh functional use breath is positively associated with sales and overall  
[06:21:57] performance. The same holds for uh operational investment breath and worker test use breath but effects are smaller for these two indices. Overall the conclusions that broader AI integration seems to be associated with higher  
[06:22:12] performance and sales. On the employment size results are a little more interesting. uh functional use breath and the operational investment breath are both positively associated with  
[06:22:24] employment decreases but worker task use breath is not. So uh of course there's no causal association here. We didn't attempt to identify any channels or mechanisms. But one interpret  
[06:22:38] interpretation of this finding is that broader functional use and operational investment may capture top-down structural redesign aimed at capital labor substitution while worker task use may operate as a productivity output  
[06:22:52] enhancer without replacing workers. And this is consistent with the augmentation high augmentation rates we we observed earlier. So in the remaining time I'm just going to put up this slide if you  
[06:23:04] will. Um so I'm just going to conclude here with the CA you know key takeaways from the paper so far. Uh growing but limited diffusion uh size matters. uh there is sectoral concentration and the  
[06:23:18] interaction between size and sector is also relevant in understanding uh the diffusion of AI. Uh we documented nar narrow functional use uh uh high use rates in marketing strategy it  
[06:23:32] comprehensive functional integration is rare most firms are minimal adopters but specialized use is also relevant. uh task augmentation is dominant, employment decreases are rare and broader structural integration  
[06:23:45] correlated with uh better outcomes and employment decreases.  
[06:23:51] >> That's all. >> Uh the next paper is the household impact of generative AI evidence from internet  

## Discussant remarks (06:47:54 – 07:02:47)

*Shared across the three papers in this session; Martin Beraja discussed all three.*

[06:48:03] >> Hi everyone. Yes. Uh thanks Eric and Karen for inviting me to discuss these uh three papers on AI productivity and economic value. And uh so know discussing three papers is kind of hard in only 15  
[06:48:18] minutes. I'm not going to go deep. I saw my role as a discussant as know framing a bit what these how these papers fit together and uh and in fact and also sort of giving you a uh know hitting the call of Matt Jackson's this this morning  
[06:48:32] on you know there's one shallow view of this p of this session and also I think it applies to other sessions which is oh these are just three papers that measure AI use and value uh at different layers national accounts firms and households  
[06:48:47] and in a bunch of other sessions we saw the same idea A that no people are just trying to paint pictures of what is going on with AI use adoption uh value and I think the call of MAT which I wholeart wholeheartedly agree on is that  
[06:49:01] we need a bit more theory to kind of see what like what are we learning from all these different pieces of evidence and the theory that I'm going to put on the table right now is that I think these papers are pointing to something deeper.  
[06:49:14] It's not just different layers. They're pointing uh to thinking about AI not just as an as a technology that uh will automate production tasks like a fancy robot but more so that is a technology  
[06:49:28] that can accelerate learning across different agents. So firms, workers and households when they learn and codify and act on on information. Okay. And and so the the the question the organizing question for the discussion is not going  
[06:49:42] to be is AI being adopted which can be you know read as a bit shallow but does AI change how agents learn and use uh information and uh and I would say the the different papers measure different  
[06:49:55] parts of this learning process as it happens through the economy starting with households searching uh ideas and jobs and learning skills and then turning into firms and workers who then qualify some sort of knowledge into organizational capital that then  
[06:50:09] aggregates into something that we can measure in the in the in the national accounts and so know starting with the last paper that here comes first is the household's paper uh where there you know their language is that no judgment is mostly used for productive online  
[06:50:23] tasks and and I like to translate that into being about pre-market entry learning in the in the following way. So one which they talk about quite a bit in the paper and another one that I which they talk more about. I think you have  
[06:50:37] the the data there to to do so. So the the first part is is is learning uh by workers or potential workers both through education and job search different informationational activities  
[06:50:49] that end up bu being built into what we will call just human capital matches across firms and workers and and know in the in the paper they they go the they they go a long way uh trying to to look at those type of activities. There's a  
[06:51:02] separate activity that also happens at home which is someone just ideating a new project, a new firm, a startup and uh and in that process no they're doing a lot of informationational research and discarding ideas that may be good  
[06:51:17] discarding bad ideas and and focusing on on on better ideas and that all fits into entrepreneurial pre-entry signals that eventually become firms. So I would really like I I'll show you later. I think this second part is actually a big  
[06:51:31] deal. it could be a big deal quantitatively in in how it fits into aggregate GDP. Uh the paper then that goes and now know we're moving within firms and uh and here I really liked in  
[06:51:45] this paper how they separated the the functions and not just saying oh these are the tasks that are performed within the firm versus not based on own it but functions that really uh align in my mind better with what is organizational  
[06:51:58] learning. So the bottom up channel that that that was discussed this idea that there's workers that are experimenting with both the tool and the task and are learning in the process of performing those tasks how to uh performing better  
[06:52:10] and which eventually leads to some codification coming from the top of those uh activities as being firm policies or broader functional uh deployment. So  
[06:52:23] again this this this this chain from individual learning to organizational capital is kind of what is what is being uh shown in in in the in the in the paper on firms and you know an initial  
[06:52:36] reaction in that paper is well if you see some of the numbers actually they look small in terms of adoption uh so far so narrow use but I'm I think that the way I read this is just early evidence for organizational learning the firms are still figuring it out exactly  
[06:52:51] how is it that we're going to uh this uh these new AI tools and uh and again I would commend the authors to really like push this idea of how is it that we move from one layer to the to to the other layer. I thought that was the most interesting part of of the paper,  
[06:53:06] not just oh here's another survey where we're measuring AI use within firms. Um lastly the the paper on on intangibles or AI organizational capital in the argate. I love this phrase. I didn't  
[06:53:20] know it from Mor it was from Mgrim and Roberts that that one way to think of organizational capital is as the accumulated stock of solutions to the meta task coordination problem. I think that's a fantastic phrase. It's a fantastic way of thinking about this and  
[06:53:33] that what AI can do here is really reduce the unique costs of coordinating activities uh within the company that then leads to organizational capital and things that we measure eventually as higher uh productivity. And so this is  
[06:53:47] really the bridge from task exposure measures to organizational learning. How how we how we gonna uh coordinate those uh those tasks. Now for those of you who have never thought about intangibles but  
[06:54:00] know a bit about physical capital and the perpetual inventory method, let me spend like two sec two minutes on on on why is it what Carl was doing hard and and and why I give her a lot of credit for trying to do this. So if you think  
[06:54:13] about physical capital, how is it that we measure it? don't go out and count machines. That's not what we do. What we do is we count the flows. So we look at say machine purchases the flows we some sort of deflator for for those prices  
[06:54:28] and then we come up with a measure of the quantity of machines that were uh added to the stock and then we accumulate those and hope that over time know the old machines depreciated enough that they don't matter anymore and we get a measure of the stock. That is the  
[06:54:41] perpetual inventory method. This works well because what you're measuring on the investment side machines is exactly that what goes into the stock. Okay. The problem with organizational capital  
[06:54:55] intangibles now the new version being AI based organizational capital is that what you're seeing say use of AI or expenditures in AI in cloud compute whatever it is like give me your your  
[06:55:08] preferred measure is not really what goes into the stock. It's an input into what goes into stock. So you're not measuring the the investment flow. Well, in particular, you can think for example that the way to build the stock of organizational capital is not just with  
[06:55:23] compute or tokens or AI use expenditures. You need to combine that somehow with say managerial and worker time. Okay? And that by combining those two things they figured out how to say produce better, coordinate those meta  
[06:55:36] tasks. That's the knowledge that is being produced. And what you would like to measure is this knowledge flow not the use of AI and that is hard because we don't observe that other part we don't observe the managerial time we don't observe the the worker time or any  
[06:55:50] other resource that goes into the the investment of uh uh that leads to organizational capital accumulation and and so what are the risks uh or when is it that this works well so suppose that you know the imagine that the investment technology is just CRS. So here you can  
[06:56:04] see very clearly that that if you have and no they did a great job at trying to measure the the decline in in the price of of of uh of AI. If you have AI that is getting cheaper then what's going to  
[06:56:18] happen within the companies that no we're going to use a bit more AI and we're going to substitute away from other things that we were using before to build organizational capital. So if we just count and accumulate all the expenditures of AI, we're gonna be  
[06:56:33] overstating the actual increase in investment because we're not seeing all the substitution that is happening say with manager time, worker time, h whatever it is. Uh so no, this only works it works perfectly when the two are perfect compliments. So if the  
[06:56:47] managerial time and AI are perfect complements, let's say to build organizational capital, then you're getting the right measure. But otherwise, you're you're overstating it.  
[06:56:55] uh so one idea here is to use you know how do you fix this so you could use substitution elasticities and I don't know if you try this for other form other forms of IT related organizational investment that we have seen in the past  
[06:57:09] and no the extrapolation would be well maybe the institution are similar between computers and managial time than they were than they are now for AI and and managial time and then we then we move we move forward uh now the second  
[06:57:23] one I this one really I don't know what to do. I think it's a it's a it's not just a problem here. It's just a more general problem that when you're measuring AI use or when you're measuring AI investments uh and  
[06:57:36] expenditures, what you're seeing is not really the part that goes into that is really investment. You're seeing both the production and the investment side.  
[06:57:47] So imagine that you know you're spending X number of tokens and that's what we see. That's what you see for each firm.  
[06:57:53] We don't know where those tokens were used for writing emails which would be like a production task or they were used for you know ideating new products or codifying information anything that looks more like a like an investment. So  
[06:58:06] again, I think this is a call for uh perfect this is a call for you know kind of trying to separate better what is investment what what part of what we're seeing as being AI use and AI expenditure is really an investment  
[06:58:20] versus something that goes into uh production and and it's maybe automating or augmenting uh those tasks. So here I think the two papers the firm and the intangibles paper can kind of like talk to each other a bit and see whether from  
[06:58:34] the surveys we can get some idea of how much of the use is really something that looks more like investment versus something that looks more like production and with that we can do maybe some some adjustments. Um uh  
[06:58:49] good so let me just close you know again going back to to theory uh a bit on what are the no I talked about learning but what are really the two mechanism of learning that can be important here. So one that the firm paper and the national  
[06:59:02] accounts paper point to is really uh learning through firm organizational capital accumulation and uh and and that may that that that may an important part of the story might only be half the story. A more important part or as  
[06:59:17] important part that I think you know receives much less attention is whether AI can help you know potential entrepreneurs or even companies select better which projects are more likely to  
[06:59:30] be successful and households the household paper is going some way into trying to measure that. I would push you to do way more on this because I think there's we we need much more work on figuring out uh this part so that AI may not only matter by speeding up our  
[06:59:45] capital accumulation but also by improving uh selection of projects in the in the economy and and so let me just you know give you you might say well why do we care about all these things is this a big deal or not and another way of asking that question is  
[06:59:59] the question that we asked in this paper with with Edward Talamas on if we had a technology that could accelerate learning like AI high then what would the aggregate uh GDP gains uh be are those potentially large or not if because if they're not large then we  
[07:00:13] could just know close shop and just think about AI as a robot h but if they are large then I think it requires much more careful thinking and measurement about uh those two those two channels so in this paper we know we came up with with this metric vault which is the value of organizational learning  
[07:00:27] technologies that and we show that in a very large class of firm dynamics models where there is learning happening you can come up with uh the gains just by looking at two sufficient statistics that I try to capture how much learning  
[07:00:41] is going on in the economy. So one is if you see an economy where mature firms are much larger than than than young firms that is an economy where organizational capital accumulation is really slow. It takes time and and it's  
[07:00:53] important and and and another the other stat statistic is if you're seeing an economy where young firms exit much more frequently than all than older firms that is an economy where selection at entry before entry is not very good. So,  
[07:01:08] we're not doing a pretty a very good job at the economy is not doing a very good job at selecting uh uh the the right projects uh pre-entry and and when you compute those statistics in in the in the in census data, you come up with numbers that are you know older firms  
[07:01:23] are three times as large as younger firms and they die uh no seven times less than than uh than uh than younger firms. And so that gives you a a gain in terms of aggregate GD potential gain of  
[07:01:36] accelerating learning if you go to the very top firms. So this accelerating a lot 40 years. So we think about this as as an upper bound of two meaning that no accelerated learning could double GDP in the US. And um now what we don't know in  
[07:01:50] this paper with the dart is well how much does AI accelerate learning? We can look at different scenarios. Imagine you can accelerate learning by 40 years, 30, 10, two and then the gains are very different whether you have you know  
[07:02:04] whether a accelerates learning by many many years or not and also depending on the mechanism that you have in mind whether it works through mainly through organizational capital or through the better selection and and prediction of projects. So then the call for all these  
[07:02:18] you know the the papers that are working on measurement is try to frame the discussion more in terms of these mechanisms and and and and the efficiency gains in each of these mechanism in terms of for examples years accelerated uh of learning because I  
[07:02:32] think then that's going to create much more value for the rest of us that are trying to quantify uh sag gains or or macro implications of this evidence to try to map them better to to to the models there. Okay. Thank you very much.  

## General Q&A (07:02:47 – 07:10:06)

*Shared across the three papers in this session.*

[07:02:47] >> [applause] >> We've got our microphone up there if people want to line up if you have questions or comments.  
[07:03:00] >> All right. Yeah, please go ahead. >> Yeah, thank you. Um, so thank you um all for coming and presenting uh to us. I think these were amazing papers. Um, I guess I have a question maybe for the first two especially about AI diffusion.  
[07:03:15] Um so there was a lot of focus on the AI diffusion maybe within workers workers adopting AI and within organizations where um either from the top down or bottom up it's used more but if we think  
[07:03:28] about AI diffusion across like an entire industry or across the economy there's also maybe similar to the idea of a selection effect uh mentioned by the discussant um you could have new firms who pop up who are early adopters of AI  
[07:03:42] technology. I mean, I'm I'm just struck by we're um in Silicon Valley right now, the home of startups, and there are so many uh AI native firms hoping to disrupt, you know, existing incumbents and existing uh industries um with uh  
[07:03:55] some sort of solution that's fully AI uh based. Um so I was wondering um if there are any thoughts on like metrics uh that could be used to track this to see if it's important or not. And I guess if  
[07:04:07] you have any personal opinion about um this uh avenue being important for AI diffusion within the economy. Thank you.  
[07:04:18] >> Do any of the Yeah. Uh could you get a micro There's a microphone right there.  
[07:04:27] >> Is it like this maybe? Uh okay, it works. So yeah, that's a great question.  
[07:04:33] One thing we haven't so far accomplished in this paper is that we didn't match the BTOS data to LBD which is the ne next task and it's almost done and then we're going to be able to bring in the  
[07:04:45] age variable for firms right so then we can really look at the diffusion patterns for young firms versus old firms and you know basically uh you know chart out the discrepancies there and  
[07:04:58] see if there's a um extra propensity for younger firms to come in with already AI technologies for example and I do have a related paper that looks at um business formation micro data from the business  
[07:05:11] formation statistics of census bureau where uh using text analysis I'm able to sort of like fuzzily identify uh which firms are AI related uh which uh business applications are AI related and  
[07:05:26] aiming to start AI related businesses. So we have seen an acceleration uh in AI rel related business applications and formations in the last uh eight years since the you know deep learning revolution trying to commercialize these  
[07:05:41] technologies. So definitely there is some evidence of that you know the diffusion through entry and new firm formation is definitely there and maybe we could do more to quantify that. I  
[07:05:53] agree. Yeah. >> Carol, do you have anything to No, I don't have anything to add. I I to that I think uh the the linking to the age  
[07:06:06] variable uh is you know answers the question the gentle the the gentleman had. Um I I think that one does have to bear in mind if somebody is using AI  
[07:06:17] technology to disrupt a traditional business as opposed to an AI firm that's operating in that AI stack. I mean they're just serving very different roles in the economy. Both both are  
[07:06:32] good. Um but um it's the diffusion is the the first example that I used.  
[07:06:41] >> Terrific. Eric, are you go? >> Yeah, it's a question about the com score and maybe I missed it, but the adoption was, you know, if they'd ever they visited chat the website and then there were adopters there ever after,  
[07:06:55] but it seems like the the power of the comore data if you have this second by second. I think most people use or a lot of people anyways use chatpt through the website and wouldn't that give you a very detailed information about  
[07:07:09] extensive use and people who use it for a while and then stop use it. I've seen some of the adoption surveys see falling usage by by some people or or just level of intensity and I don't know if you could do more with that or if I missed a  
[07:07:22] reason why you weren't >> Yeah. No, I mean I think it's a very welltaken point. Um we have the intensive margin based measure of just you know in a given quarter how many seconds or what share of your browsing  
[07:07:35] seconds did you spend on chatbt or openai.com?  
[07:07:40] uh but it's not something that we use to kind of uh extend any of the the insights of of our analyses. It was more just kind of as a you know robustness check for the indogenous variable. But I agree with your point.  
[07:07:52] >> You can say I mean honestly that would have been the first thing I would have looked at. But but what what what is is there anything you can say about what what that showed? I mean >> yeah I mean we find similar treatment effects essentially when we scale it by uh that alternative endogenous variable.  
[07:08:04] But I think kind of to your you know more to what you're asking I think that you know there's a lot more that we can push on in terms of things like is there a generative AI divide when it comes to if you start using it do you keep using it like is the divide we see the  
[07:08:19] >> John Hartley found some evidence in their their survey that that usage was going down I don't know whether it was like a blip or noise or whether it was real did you see anything like that >> um I honestly do not think that we have  
[07:08:33] No. So we see that the intensive margin is >> usually end before when when does your data end? 2020.  
[07:08:39] >> So it's about to be extended up until the end of 2025, but currently it ends in December 24.  
[07:08:46] >> Well, no, we haven't seen any evidence of a dip so far, but I think within person that could, you know, it could be masking within person heterogeneity. So yeah, >> Eric, that's great. Just to add to that um so we are also exploiting the  
[07:09:00] substitution between computer versus mobile because uh when we talk to the open AI I think you know there's an important fact folks are here um so uh there is a kind of a massive rise of  
[07:09:13] mobile use at the end of our sample I think it's a June of 2025 I believe so that's that's another thing I think you're mentioning we're we're really excited to look into. Yeah.  
[07:09:27] >> Okay. Since we have about a minute left, I just want to ask if any of the paper authors have anything to say in reaction to Martin's excellent discussion.  
[07:09:45] [laughter] >> All right.  
[07:09:49] >> I just want to thank you. It was really nice and you gave us some new ideas to exploit entrepreneurship. Um, yeah, that's cool.  
[07:09:57] >> Okay, we have a 15minute break, a little more networking time, and then uh please be back here promptly at uh 4 o'clock and we have an amazing panel. See you then.