No Measurable Productivity Boost from Generative AI?

With the caveat that knowledge worker and office worker “productivity” is difficult to impossible to measure, a study by ActivTrak was unable to find evidence of productivity improvements in 2023 due to use of generative artificial intelligence, primarily ChatGPT.

The study suggests that overall productivity increased by about eight minutes from the beginning of 2023 to the end, from all sources. Roughly 22 percent of workers across industries studied used ChatGPT or generative AI. 

  

source: ActivTrak

But the study does not specifically try to evaluate the productivity impact of generative AI. That should not be surprising. 

Though some observers believe AI could produce a startling rate of productivity increase perhaps an order of magnitude higher than what we have experienced in recent decades, there are reasons to believe such forecasts are far too optimistic. 

Knowledge worker and office worker productivity--aside from being tough to measure--has not historically shown high impact from information technology adoption, despite our sense that this should be the outcome. 

Since 1970, the actual impact of IT arguably is rather slight. Productivity growth from all sources--including IT--is on the order of 0.5 percent up to about 1.5 percent per year, maximum, from all sources, including personal computers, the internet, cloud computing, mobile devices, open source and everything else in the information technology field, plus all other influences. 

Industry Sector (NAICS Code)	Description	Percent Change in Labor Productivity (1970-2022)
Total Nonfarm Business (All)	Covers all industries except agriculture, government, and private households	1.0%
GoodsProducing Industries (1133)	Includes mining, construction, and manufacturing	0.4%
Manufacturing (3133)	Factory production of goods	0.5%
Construction (23)	Building, renovation, and maintenance of structures	1.2%
Mining (21)	Extraction of minerals and natural resources	2.3%
ServiceProviding Industries (4892)	Covers a wide range of service businesses	1.4%
Wholesale Trade (42)	Selling goods to businesses in bulk	1.2%
Retail Trade (4445)	Selling goods directly to consumers	0.4%
Transportation and Warehousing (4849)	Moving people and goods	2.1%
Information (51)	Publishing, broadcasting, and telecommunications	2.5%
Financial Activities (52)	Banking, insurance, and real estate	1.2%
Professional and Business Services (5456)	Legal, accounting, consulting, and scientific services	2.0%
Education and Health Services (6162)	Schools, hospitals, and other social services	1.3%
Leisure and Hospitality (7172)	Accommodation, food services, and entertainment	3.9%
Other Services (8189)	Repair shops, personal care services, and religious organizations	0.8%

But impact is likely to be uneven. Keep in mind that GenAI is used to create content. So it is firms in industries that principally “create content” that stand to benefit the most. 

Recent strikes by Hollywood actors and writers illustrate that point. The entertainment media industry--which principally creates content--is among those most exposed to GenAI and most able to use the tools. 

Media also is an industry estimated to have had higher productivity gains in 2022 than most others. Where all “non-farm” industries might have seen an average one-percent productivity gain in 2022, media saw a boost of up to 2.5 percent, according to Bureau of Labor Statistics figures. 

Likewise, firms in the product design business; marketing and advertising; pharmaceutical development and finance industries are among segments that routinely create content as a core function, and might therefore be expected to be areas where GenAI has early impact. 

Professional services might have seen a boost in 2022 productivity of perhaps two percent. 

On the other hand, many industries do not rely principally or even significantly on content creation, and should lag in terms of GenAI producing measurable business results. 

Basic resource extraction, such as mining or forestry should see limited impact, one way or the other. In fact, goods-producing industries tended to see negative productivity growth in 2022. AI might help, but we need to be realistic about the degree of potential change.

Is Service Provider "Core Competency" Changing?

If you ask just about any telco executive or middle manager what their firm’s “core competency” is, the traditional answer almost always has something to do with “creating and operating communications networks.”

Keep in mind that a core competency is not a list of “things we do well.” Instead, it is the combination of resources and skills that give a company a strategic advantage in the marketplace. It's essentially what a company does best and what differentiates it from competitors.

In the monopoly era before 1980, that answer would have made sense, as the law prohibited all others from operating in the business. In the competitive era the traditional answer is eroding. Firms such as Google and others actually fund, build and operate their own connectivity networks. So do cable TV companies and all sorts of mobile communications companies. 

So it is hard to make the argument that “creating and operating networks” actually is the core competency anymore, especially as new suppliers continue to enter the market.  

Decade	Core Competency Claim	Supporting Studies	Limitations
1980s	Network Building & Operation	* Caves, R. E. (1982). Multinational enterprise and economic structure. North-Holland ([study of vertical integration in telecoms])	Network infrastructure was seen as a significant barrier to entry, giving telcos a strong advantage.
1990s	Network Building & Operation (contested)	* Faulkenberry, G. D., & Caves, R. E. (1995). Telecommunications infrastructure: The missing link to economic development. Brookings Institution Press. ([study on telecom infrastructure's role in economic development])	Technological advancements and regulatory changes began to challenge the dominance of network ownership.
2000s	Network Building & Operation (further challenged)	* Witt, R. (2004). How the mobile network became ubiquitous: The telco industry and the regulators. Information, Communication & Society, 7(1), 73-95. ([analysis of co-opetition and network sharing in mobile telecoms])	The rise of wholesale fiber networks and joint ventures cast doubt on network ownership as the sole core competency.
2010s - Present	Shifting Landscape	* Brynjolfsson, E., Rockaway, R., & Van Alstyne, M. W. (2017. Platform revolution: How networked markets are transforming the economy and making winner-take-all competition irrelevant. W. W. Norton & Company. ([study on platform business models in telecoms])	Regulatory expertise, innovation & service delivery, and customer experience are increasingly seen as crucial alongside network capabilities.

Precisely where core competency might eventually be identified is among the core questions for industry leaders, especially as “network operation” ceases to be a clear and indisputable core competence. 

Some might suggest the eventual core competency could be as an “orchestrator of connectivity.” In other words, some tier-one telcos could evolve as system integrators for global communications across all networks (fiber, wireless, satellite) for businesses and consumers. They might own some of the resources used, but largely function as one-stop-ship global connectivity suppliers. 

Perhaps some might try to become “platforms” and “enablers,” essentially becoming connectivity infrastructure providers supporting business partners who develop and deliver the actual end-user services.

Roles for smaller firms could be dramatically different, as smaller telcos might not have the means to support global integrator roles, and might function as suppliers of local resources in a particular geography to the larger tier-one suppliers. 

Arguably less likely are evolutions that would reposition telcos as suppliers of targeted advertising, network optimization solutions, or location-based services supported by their ability to target network users. 

And though cybersecurity is increasingly embedded into all hardware and software, it seems unlikely that most tier-one telcos and smaller firms can become cybersecurity specialists.

The core competency underlying all  these future scenarios includes ability to manage and integrate complex communication ecosystems; creating seamless and personalized communication experiences.

If I had to guess right now, I’d assume the global connectivity integrator role would make sense for a handful of tier-one providers. 

The enabler role might become more prominent for smaller service providers, in the same way that the internet separates app development and ownership  from network access; or compute infrastructure from apps. 

The big switch is from specialized app provider (voice and data connectivity for business customers) to “internet access and transport” provider; global system integration rather than regional franchise; retail point of contact rather than physical layer transport provider. 

Even the move to joint ventures, wholesale network operation and ability of third parties to enter the business all suggest that the “network operator” competence is changing to something else.

The issue is how firms discover that new core competence. 

FTTH Payback Depends on the Business Model

It always is difficult to figure out the payback period for any fiber-to-home investment, as the number of key competitors; regulatory encouragement or discouragement issues; retail versus wholesale business models and possibility of joint venture arrangements; amount of aerial versus underground construction vary widely. 

But everyone agrees the FTTH asset is long lived, and has payback periods that are optimistically less than 10 years, but can run up to 20 years. And that obviously increases incentives for FTTH investors to reduce risk and lessen capital investment burdens. 

Joint ventures that reduce any single firm’s capital investment and also speed up deployment rates are an obvious solution. So are wholesale business models that increase network utilization rates. 

Study Title	Source	Market	Payback Period Estimate
"Fiber to the Home: The Economic Viability for Competitive Local Exchange Carriers"	National Telecommunications and Information Administration (NTIA), U.S.	U.S.	7 to 14 years
"The Economics of Full Fibre Deployment: Costs and Benefits"	University of Oxford, U.K.	U.K.	10 to 20 years
"Fiber to the Home: The Path to Success"	Federal Communications Commission (FCC), U.S.	U.S.	5 to 15 years

Firms backed by private equity have different business models than other long-term operators of connectivity assets. PE-backed firms aim to create value (typically double the asset value within seven years) and then sell the assets. 

That is a different payback model than used by connectivity service providers who operate for the long term, where fundamental issues of free cash flow, revenue growth and profit, as well as the ability to pay dividends, are the key constraints. 

And so it is with investors in fiber-to-home assets. For a private equity firm, all that matters is the ability to flip the asset for a higher multiple than acquired assets, even after the FTTH upgrades.

For an operator of internet access services, payback must include considerations of revenue per customer and revenue per passing as well as profit margins long term.

Asset value drives the logic of investment for a PE firm. Gross revenue, operating cash flow and ultimately profits are what matter for a service provider. But acquiring assets from PE and other firms can make sense for a service provider if it accelerates time to market and time to cash flow, while also possibly having the benefit of eliminating a potential key competitor in a market.

Are Massive Energy Consumption Forecasts for AI Data Center Operations Correct?

With the caveat that the following chart shows both data center energy consumption as well as cryptocurrency mining operations, manufacturing and electrification of fossil fuel operations, energy demand fueled in part by artificial intelligence is going to double consumption within six or seven years.

How that fits with pledges of net zero carbon dioxide footprints is unclear to some of us. 

source: Financial Times 

That noted, we cannot ever rule out human creativity and technology advances such as ways to dramatically lower energy consumption for AI-related operations, but some researchers suggest the demand for data center energy could grow much more than a doubling over the next six years. 

Study Title	Publisher	Year	Projected Growth
"Power Consumption of Large Scale Data Centers:"	Institute of Electrical and Electronics Engineers (IEEE)	2019	Up to 10x growth by 2030
"The State of AI 2022	Stanford University	2022	Doubling of training compute every 3-4 months
"AI and the Environmental Footprint of Computing”	Cornell University	2020	Up to  1000x growth
"The AI Revolution's Power Squeeze"	Nature: https://www.nature.com/	2020	Up to 3x by 2025
"How AI is Changing the Data Center"	McKinsey & Company: https://www.mckinsey.com/	2022	2.5x to 4x by 2025
"The Implications of Deep Learning for Data Center Design"	Joule: https://www.sciencedirect.com/journal/joule	2019	Doubling within a few years

Most assumptions we make about future energy consumption are necessarily linear, and assume AI operations at data centers are not essentially examples of complicated ecosystems that are necessarily “chaotic” and subject to huge changes produced by small inputs that are part of our assumptions.