The U.S. Economic Outlook for AI
What does the White House say about AI’s growing influence? It’s long on potential, but adoption is lagging
“Artificial intelligence (AI) systems touch the lives of virtually every American,” said An Economic Framework for Understanding Artificial Intelligence, an overview of the economic potential of AI which was recently released by the White House Council of Economic Advisors (CEA) as part of their 2024 Economic Report of the President. “In recent years, AI systems have advanced rapidly as recent developments in computing, data availability, and machine learning models have simultaneously come together to produce rapid improvements.”
“Still, much remains unknown,” the CEA added. “While AI’s capabilities will depend in part on the technology itself, its effects will be shaped by economic, regulatory, and social pressures.
How society deploys this technology and what technology-specific guardrails are implemented will be critical factors in determining both the breadth and magnitude of its effects.”
“An economic framework, combined with a basic understanding of AI technology, allows us to make predictions about when, how, and why AI may be adopted. While such a framework can also tell us what broader effects AI adoption may have, applying economic insights to an evolving and proliferating technology like AI is especially challenging. However, it is also especially valuable, because decisions made at the onset of a new technology have a greater influence on its eventual impact.”
The report concludes that AI technology “changes have the capacity to benefit everyone, but
recent empirical evidence shows that broad-based benefits are not guaranteed.
Sensible policies to encourage responsible innovation, protect consumers, empower workers, encourage competition, and help affected workers adjust are critical.”
The AI Economic Framework is a rather long and comprehensive document. Let me briefly discuss a few of the its key points.
Prediction is Improving but Faces Constraints
In their various papers and books, University of Toronto professors Ajay Agrawal, Joshua Gans, and Avi Goldfarb argued that the best way to assess the economic impact of a transformative technology is to look at how the technology reduces the cost of a widely used function. Computers, for example, have dramatically reduced the cost of digital operations like arithmetic and logic by several orders of magnitude.
As a result, we’ve learned to define all kinds of tasks in terms of digital operations, e.g., scientific research, financial transactions, inventory management, word processing, photography. Similarly, the internet has reduced the cost of communications and the Web has reduced the cost of access to information, which has led to a huge increase in applications based on communications and information, like music and video streaming, and digital media.
Current AI systems are quite different from classic digital technologies. Instead of being explicitly programmed, AI systems are trained with large amounts of data, which are then used to make predictions using sophisticated statistical methods and powerful computational techniques.
AI is essentially a prediction technology, whose key economic impact is to reduce the cost and expand the number and variety of applications that rely on predictions.
The more data used in their training, the more advanced the algorithms used to analyze the data, and the more powerful the computers, the better the AI predictions. “The ability to make predictions often allows improved decision-making, even in the face of uncertainty. As a result, AI systems can automate more tasks than prior technologies and improve the work quality of existing processes.”
Garbage In, Garbage Out
The CEA reminds us that data plays a central role in AI. “Data are key informational inputs into AI systems, and they are central to the way AI performs. AI systems make informed predictions because they use the correlations embedded in data. Many different changes have contributed to improvements in AI systems, including improvements in algorithms and increased availability of computational resources. Nonetheless, developers of AI-based prediction models continue to grapple with many of the same data-related challenges that statisticians and econometricians have faced for decades. To understand AI technology as a whole, it is helpful to understand the unique role that data and data-related constraints play.”
“Prediction models typically perform well in situations that look much like the data they are trained on. In contrast, rare or novel circumstances where the past is a poor guide to the future make prediction more challenging, as do data limitations that might not immediately be apparent.”
Adoption is Difficult and Invariably Lags the Technological Frontier
Artificial intelligence has been a promising area of computer science research since the mid-1950s, with the aim to develop intelligent machines capable of handling a variety of human-like tasks. But after decades of promises and hype, in the 1980s it went through a so called AI winter of reduced interest and funding that nearly killed the field. AI was reborn as a data-centric discipline over just the past couple of decades with innovations in big data, predictive analytics, and machine learning algorithms. Some of the most impressive advances in AI, large language models (LLMs) and generative chatbots, are quite recent, still in their very early years.
AI has finally emerged as a historically transformative technology which, over time, will radically change the economic environment. But, as we’ve learned over the past two centuries,
realizing the potential of historically transformative technologies requires large intangible and often unmeasured investments and a fundamental rethinking of the organization of production itself.
As a result, there’s generally been a significant time lag between the broad acceptance of a historically transformative technology and its ensuing impact on companies, industries, and individuals.
“The markets for AI are already adapting, with investment and start-up activity both increasing in recent years,” notes the CEA report. “Businesses specializing in cloud computing and AI deployment have also since emerged, lowering costs and expanding adoption.” However, AI faces a variety of potential impediments to widespread adoption.
For example, “even when data are available to train an AI system, there may be additional data-related constraints on adoption. Many firms may not yet collect the necessary data for certain AI implementations, and they may face substantial challenges in beginning to do so. In other cases, systems do not receive feedback sufficient to judge the quality of their own predictions after they have been made. Finally, even when the data exist, legal restrictions like copyright may prevent their use.”
AI Has the Potential to Be Even More Transformative in the Future
AI is a general purpose technology, like electricity, computers, and the internet, capable of supporting a wide variety of applications. These technologies improve over time and lead a number of major complementary inventions.
The economic impact of AI is likely to be significantly larger and more wider-reaching than the initial use cases would suggest.
“While some services have been redesigned on the basis of AI, and some new technologies have been built with AI from the ground up, many systems and processes that could be redesigned to take advantage of AI have not yet been updated. Firms that invest in AI are showing signs of increased product innovation, but they do not yet show evidence of process innovations that might arise from a more thorough restructuring of their operations.”
AI and Labor Markets
“What does AI’s ability to undertake tasks previously performed by humans mean for labor and the labor market? On net, will AI complement workers, yielding increased jobs, productivity, and prosperity? Or will prediction models substitute for human labor, yielding a world where fewer people are needed to work, but also where fewer people can contribute to the economy while also earning a living?”
Throughout the Industrial Revolution there were periodic panics about the impact of automation on jobs, going back to the Luddites, — textile workers who in the 1810s smashed the new machines that were threatening their jobs. Automation fears have understandbly accelerated in recent years, as our increasingly smart machines have been applied to activities requiring intelligence and cognitive capabilities that not long ago were viewed as the exclusive domain of humans.
But each time those fears arose in the past, technology advances ended up creating more jobs than they destroyed. While automation does indeed substitute for labor, automation also complements labor. Automating the more routine parts of a job will often increase the productivity, earnings and quality of workers, by complementing their human skills with new technologies that automate some of the tasks, enabling them to focus on those aspect of the job that most need their attention.
“A recent Gallup poll found that 75% of U.S. adults believe AI will lead to fewer jobs,” wrote MIT economist David Autor in a recent article, “AI Could Actually Help Rebuild The Middle Class.” “But this fear is misplaced. The industrialized world is awash in jobs, and it’s going to stay that way. Four years after the Covid pandemic’s onset, the U.S. unemployment rate has fallen back to its pre-Covid nadir while total employment has risen to nearly three million above its pre-Covid peak.”
“Due to plummeting birth rates and a cratering labor force, a comparable labor shortage is unfolding across the industrialized world (including in China),” he added. “This is not a prediction, it’s a demographic fact. All the people who will turn 30 in the year 2053 have already been born and we cannot make more of them.
Barring a massive change in immigration policy, the U.S. and other rich countries will run out of workers before we run out of jobs.”
Conclusions
“AI has the potential to increase economic well-being,” concludes the White House Council of Economic Advisors. “Like many previous technologies, it will do so by transforming the economy in both expected and unexpected ways. Economic theory demonstrates that the changes have the capacity to benefit everyone, but recent empirical evidence shows that broad-based benefits are not guaranteed. Sensible policies to encourage responsible innovation, protect consumers, empower workers, encourage competition, and help affected workers adjust are critical.”
“The future path of technological change is always uncertain,” the CEA added. “As AI’s role in the economy grows, the Federal Government will need to continually evaluate its institutional framework. Only by thinking broadly about AI and its effects can society balance the technology’s potential for harm against its many possible benefits.”
This blog first appeared May 16 here.
