If AI Emerges as a General-Purpose Technology, Watch for Both Disruption and Creation

Any general-purpose technology might be envisioned as a set of layers of other technologies that build on it. Many could agree that GPTs are characterized by pervasiveness, flexibility, spillover effects and transformative impact. 

So the internet might underpin layers of core infrastructure and industries and businesses built around protocols such as TCP/IP and physical networks and industries (mobile and fixed networks, terrestrial and satellite wireless networks). 

Then there might be layers of roles and businesses supplying web technologies such as HTML, CSS, JavaScript, and related web development tools that enable building websites and web applications.

Networking technology including routers, switches, firewalls would be another layer. 

So would databases, cloud storage, and content delivery networks.

Then there would be many application and service layers for communication (e-mail, instant messaging, video conferencing, and social media platforms) e-commerce and online marketplaces, content and entertainment, social media, video and audio streaming or online gaming. 

Internet of Things businesses built around smart devices, sensors, and connected appliances, as well as many types of business software could be listed.

Some might include artificial intelligence as among the layers built on the internet. But some of us would say AI is a new general purpose technology that will create its own pyramid of technologies, businesses, industries and applications. 

Era	General Purpose Technology	Impact
Pre-Industrial	The Wheel	Revolutionized transportation, agriculture, and warfare. Led to the development of roads, carts, and other wheeled vehicles.
18th Century	The Steam Engine	Powered the Industrial Revolution, driving mechanization and mass production in factories, transportation (trains, ships), and agriculture.
19th Century	Electricity	Transformed daily life with lighting, appliances, communication (telegraph, telephone), and industrial processes.
20th Century	Internal Combustion Engine	Propelled transportation revolutions with automobiles, airplanes, and ships. Changed industries, warfare, and leisure activities.
20th Century	Electronics & Semiconductors	Enabled miniaturization of devices, leading to computers, radio, television, and countless electronic gadgets.
20th Century	The Internet	Connected the world, democratized information access, facilitated communication, and fueled e-commerce, digital services, and the knowledge economy.

And some of those roles or industries might presently be viewed as built on “internet” foundations. 

Intelligent infrastructure such as smart cities, autonomous vehicles, adaptable robotics, “personalized” healthcare, neurotechnology (brain-computer interfaces),  bionic limbs and prosthetics and much “metaverse” style immersive experiences, plus much of virtual and augmented reality, hyper-personalized content creation, AI-powered companions, precision agriculture and other use cases that might today be attributed to the  “internet” GPT might eventually be properly seen as built on AI as a GPT. 

Perhaps analogies can be seen in the Apple iPhone and Google search. Apple did not invent the smartphone or the mobile phone. But it completely reshaped the business, destroying Nokia and BlackBerry in the process as former market leaders. 

Google was not the first search engine, but it destroyed Altavista and other existing search engines in the market. The point is that many existing industries might be fundamentally reshaped if AI emerges as a GPT. 

And as has been the case before, AI might reshape and disrupt existing industries, functions and roles, in addition to spawning entirely-new industries, as all prior GPTs have done.

Will IoT Boost Productivity? How Long Will it Take?

The lag time between first deployment of a general-purpose technology (steam engine, railroad,, electricity, electronics, automation, automobile, the computer, the internet) and quantifiable productivity increases is not immediate, not clearly and unmistakably causal, and sometimes impossible to isolate from the impact of other general-purpose technologies.

That is important because we cannot determine whether important new technologies actually increase productivity--although people mostly assume it does--or not. Nor can we see with precision how long it will take: gains often take decades to appear in quantifiable form.

That is worth keeping in mind in assessing the return from internet of things, artificial intelligence, connected vehicles and so forth.

Consider the impact of electricity on agricultural productivity.

“While initial adoption offered direct benefits from 1915 to 1930, productivity grew at a faster rate beginning in 1935, as electricity, along with other inputs in the economy such as the personal automobile, enabled new, more efficient and effective ways of working,” the National Bureau of Economic Research says.  

There are at least two big problems with the “electricity caused productivity to rise” argument. The first is that other inputs also changed, so we cannot isolate any specific driver. Note that the automobile, also generally considered a general-purpose technology, also was introduced at the same time.

That is not to say correlations between important new technology and process efficiency are undetectable.

Looking only at use of machine learning, error rates in labeling the content of photos on ImageNet, a dataset of over 10 million images, have fallen from over 30 percent in 2010 to less than five percent in 2016 and most recently as low as 2.2 percent, say researchers working for NBER.

Likewise, error rates in voice recognition have decreased to 5.5 percent from 8.5 percent in 2017, for example.

At the same time, “there is little sign that they have yet affected aggregate productivity statistics,” the researchers note.  Labor productivity growth rates in a broad swath of developed economies fell in the mid-2000s and have stayed low since then.

“For example, aggregate labor productivity growth in the U.S. averaged only 1.3 percent per year from 2005 to 2016, less than half of the 2.8 percent annual growth rate sustained from 1995 to 2004,” NBER researchers say.

source: National Bureau of Economic Research

“Fully 28 of the 29 other countries for which the OECD has compiled productivity growth data saw similar decelerations,” they say. “The unweighted average annual labor productivity growth rates across these countries was 2.3 percent from 1995 to 2004 but only 1.1 percent from 2005 to 2015.”

So how do observers explain the apparent failure of big applications of technology to produce productivity gains? “False hope” is one explanation.

“The simplest possibility is that the optimism about the potential technologies is misplaced and unfounded,” NBER researchers say. Perhaps new technologies won’t be as transformative as many expect.

More compelling, perhaps, is our inability to measure the productivity gains. Many new technologies, like smartphones, online social networks, and downloadable media involve little monetary cost.

That poses an obvious challenge when only quantifiable price metrics can be used. A personal computer that costs 10 percent less, but supplies double the computing power or memory actually might be deemed a decrease in economic activity, for example.  

Technology improvements that boost qualitative power or potential utility might not show up in price metrics in a fully-capturable way, as imputed value is higher, but price lower. But we cannot measure higher possible value; only price changes.

Another argument is that the impact of potentially-transformative technologies is limited by limited diffusion (not all firms and industries use them equally well). In other words, the gains are not equally distributed. Some industries and firms seem to capture most of the benefits.

Perhaps the most-persuasive opinion is that it takes a considerable time to sufficiently harness the power of a new general-purpose technology, since whole business processed need to be created before the advantages can be reaped.

The bottom line: we assume IoT improves productivity, as we assume electricity and broadband also contribute. But we need to invest in a measured way, as the actual benefits might not show up for a decade or two.

That might be the case for new 5G-based enterprise and consumer use cases as well.

When Will AI Capex Payback Happen First?

Most of us would likely agree that artificial intelligence benefits are going to take a while to be seen almost anywhere except the financial results of infrastructure providers, who clearly will benefit. Nor would that ever be unusual when an important new technology--not to mention a possible new general-purpose technology--first emerges. 

Indeed, analysts at Goldman Sachs say “leading tech giants, other companies, and utilities to spend an estimated $1 trillion on capex in coming years, including significant investments in data centers, chips, other AI infrastructure, and the power grid.” 

Still, “this spending has little to show for it so far.” Nor would one realistically expect to see quantifiable results so early. The pattern with general-purpose technologies is that the platforms and infrastructure must be built first, before use cases and apps can be developed. 

Also, some functions are more susceptible to generative AI impact, for example, than others. 

Most of us would be willing to concede that customer service is one area where generative AI, for example, should produce results. Functions with many repeatable elements are commonly thought to be susceptible to AI automation. 

In a survey conducted for Bain, enterprise executives reported that better results were seen in sales; software development; marketing; customer service and customer onboarding, for example. Between October 2023 and February 2024, though, most other use cases seemed to produce less favorable outcomes than expected. 

source: Bain 

Generative AI thrives on well-defined patterns and processes, so jobs involving repetitive tasks with clear rules and minimal ambiguity are likely candidates for early change. 

But lots of functions and tasks are not routine or well structured; not simple but complex, so the range of use cases that can benefit near term is arguably limited. 

As the report notes, Daron Acemoglu, Institute Professor at MIT, estimates that only a quarter of AI-exposed tasks will be cost-effective to automate within the next 10 years, implying that AI will impact less than five percent of all tasks.

Most of us would be willing to concede that customer service is one area where generative AI, for example, should produce results. Functions with many repeatable elements are commonly thought to be susceptible to AI automation. Generative AI thrives on well-defined patterns and processes. Jobs involving repetitive tasks with clear rules and minimal ambiguity. 

All that noted, the first quantifiable results will be seen among suppliers of infrastructure, as apps cannot be built until the infrastructure is in place.   

GPT/Possible GPT	Infrastructure Provider	Early Revenue Gains
AI/Large Language Models	NVIDIA	171% year-over-year revenue increase in Q2 2023, driven by demand for AI chips
Internet	Cisco Systems	Revenue grew from $69 million in 1990 to $22.3 billion in 2001 as internet infrastructure expanded
Personal Computers	Intel	Revenue grew from $1.9 billion in 1985 to $33.7 billion in 2000 as PC adoption surged
Electricity	General Electric	Revenue increased from $19 million in 1892 to $1.5 billion in 1929 as electrical infrastructure spread
Railroads	Steel Companies (e.g. Carnegie Steel)	U.S. steel production grew from 68,000 tons in 1870 to 11.4 million tons in 1900

That noted, it also could be said that there has been overinvestment--at some point--in infrastructure for past general-purpose and new technologies. It also might be noted that application and device over-investment also occurs, early in the adoption of a new technology. 

Technology	Time Period	Description of Over-Investment
Railroads	1840s-1850s	Excessive railroad construction and speculation led to financial panics in 1857 and 1873 in the US and UK
Automobiles	1910s-1920s	Hundreds of car companies were founded, with most failing as the industry consolidated
Radio	1920s	Rapid proliferation of radio stations and manufacturers, followed by consolidation
Internet/Dot-com	Late 1990s	Massive speculation in internet-related companies led to the dot-com bubble and crash in 2000
Renewable Energy	2000s-2010s	Over-investment in solar panel manufacturing led to industry shakeout
Cryptocurrencies	2010s-2020s	Speculative frenzy around Bitcoin and other cryptocurrencies

But there is a difference between “over-investment” and the proliferation of would-be competitors in a new market. It always is normal to see more startups in any area of new information technology than there are surviving firms once the market is mature. 

The difference between over-investment and normal competition in a new market can be subtle. What might not be subtle is the lag time between capex investments and revenue realization, for firms not in the "picks and shovels" part of the ecosystem.

Infra suppliers already have profited.

Lenovo CIO Study Finds a "To be Expected" Assessment of AI

According to Lenovo's third annual study of global CIOs surveyed 750 leaders across 10 global markets, CIOs do not expect to see clear and positive return on investment from their artificial intelligence investments for two to three years. 

source: Lenovo 

We should not find this surprising. Consider the last generally-recognized general-purpose technology--the internet--and the lag in perceived benefits. 

Early internet technologies (1995, for example) were less mature and reliable compared to today, with slow connection speeds (dial-up internet was the consumer standard in 1995), limited functionality (the shift to multimedia web had just begun in 1995), while enterprises had to allay their  security concerns.

The internet disrupted traditional business models, so companies needed time to develop new strategies for marketing, sales, and customer service in the digital space. That took time.

Also, though it seems clear enough now, the potential applications of the internet for businesses weren't fully understood at first. Experimentation was required.

Additionally, assessing the return on investment for early internet initiatives was difficult, as firms lacked the analytics tools to quantify the impact of online marketing, e-commerce, or other internet-based activities.

Complicating matters was the widespread failure of many e-commerce startups in the dotcom bust around 2000. Since whole firms failed, benefits were zero or negative. 

Study	Publication Venue, Year	Key Findings
"Why E-Business Fails" by Andrew McAfee	Harvard Business Review, 2002	Analyzed early e-commerce ventures and found many failed to deliver on promises, highlighting the need for a strategic shift beyond simply setting up a website.
"The Productivity Paradox in Information Technology" by Erik Brynjolfsson and Lorin M. Hitt	Journal of Economic Perspectives, 1997	Examined the early years of IT adoption and the difficulty in measuring clear productivity gains initially, suggesting a time lag for realizing benefits.
"Diffusion of Internet Commerce: A Study of Knowledge Acquisition" by Sang-Pil Han, Young-Gul Kim, and Yoonkyung Kim	Journal of Electronic Commerce Research, 2003	Focused on small businesses and found that knowledge acquisition and overcoming technical challenges were crucial for successful internet adoption.
Diffusing the Dot-Com Revolution: The State of Business Transformation in the New Millennium"James C. Brancheau, Richard B. Clark, and Thomas G. Rowan	2001	This study found that many companies struggled to transform their businesses for the internet in the late 1990s, and the early benefits were primarily cost reductions rather than significant revenue growth
"Understanding Digital Marketing ROI: A Literature Review and Synthesis"Magali Ferro, Pauline Pinheiro, and David Thomas	2014	This review of research on digital marketing ROI (Return on Investment) highlights the challenges of measuring the impact of online marketing efforts, particularly in the early days when attribution models were less sophisticated.

That tends to be the case with most information technology innovations, other studies have found, looking at IT in general, e-commerce in specific or productivity. 

Study Title	Publication Venue	Date	Key Conclusions
The Elusive ROI of IT Investments	Strategic Management Journal	1997	Examined IT investments in large firms and found difficulty in directly measuring ROI (Return on Investment) due to factors like long-term strategic benefits and integration challenges.
From Bricks to Clicks: Does IT Pay Off?	Information Systems Research	2002	Analyzed data from over 200 firms and found a delayed effect of e-commerce initiatives on profitability. Early adopters often faced challenges like website development costs and changing consumer behavior.
The Productivity Paradox in Information Technology	The Review of Economic Studies	2003	Investigated the impact of IT on US productivity growth in the 1990s and found a "productivity paradox" where benefits weren't immediately apparent. The study suggests a "learning period" was needed for firms to leverage the internet effectively.
A Longitudinal Analysis of Web Site Traffic and Sales	Marketing Science	2004	Analyzed website traffic and sales data for multiple firms and found a positive correlation, but it took time for website traffic to translate into significant sales growth.
The Productivity Paradox in a Service Economy	Quarterly Journal of Economics	1998	Robert J. Gordon analyzed data from the US economy and found a productivity slowdown despite the rise of computers and the internet in the 1980s and 1990s. The study suggests a lag between technology adoption and measurable economic impact.
Diffusing the Dot-Com Revolution: An Organizational Perspective	Academy of Management Journal	2000	Andrew S. Melville, Thomas Durand, and Nina G. Guyader explored how established firms adopted e-commerce in the late 1990s. They found challenges in integrating new technologies with existing processes, leading to slow initial returns.
From Bricks to Clicks: Determinants of Success in Online Retailing	Journal of Retailing	2002	Kenneth C. Lichtenstein, James A. Lumpkin, and Elizabeth Van Wijnbergen analyzed early online retailers. They identified the need for significant investments in infrastructure and marketing before online channels became profitable.
Why E-Business Fails	Harvard Business Review	1999	Dorothy Leonard-Barton argued that many early e-commerce ventures failed due to a lack of strategic planning and a focus on technology alone, neglecting organizational change and customer experience.

The point is that, of course it will take some time for CIOs to demonstrate meaningful outcomes from applied AI. That is always the case when an important new technology--to say nothing of a general-purpose technology, is introduced. 

Whole business processes have to be redesigned, generally speaking, before the innovations can work their magic and produce measurable outcomes.