Productivity shocks improve efficiencies, reduce costs and increase overall welfare.
Key takeaways
But that is happening with the arrival of generative artificial intelligence.
We examine the three pillars of artificial intelligence—machine learning, generative AI and artificial general intelligence
Positive productivity shocks improve efficiencies, reduce business costs and increase overall economic welfare. They redistribute wealth, creating new economic ecosystems that profoundly change business dynamics.
Given the pace of technological change over the past 30 years, the American commercial community seems well positioned to understand and capitalize on such changes. Yet the nature, breadth and depth of the changes likely to be unleashed by artificial intelligence have stirred economic unease.
To better understand what is happening, consider the 1990s technology boom. Under technological conditions that now seem primitive, productivity-enhancing capital expenditures increased by roughly 50%. Then think of the surge in productivity that followed, and it’s difficult not to get excited about the AI revolution.
Understanding what AI can bring to businesses is critical. What we call the three pillars of artificial intelligence—machine learning, generative AI and artificial general intelligence—will have profound implications for businesses and the economy.
We think artificial intelligence will change the world the way the Renaissance did 400 years ago. Gaining a better understanding of how this will happen is a worthwhile endeavor.
Working definition
Our working definition of artificial intelligence is the capacity of computers or other machines to exhibit or simulate intelligent behavior. More to the point, it’s a system that can make predictions, recommendations or decisions influencing real or virtual environments given a set of objectives.
That definition indicates that business operations will gain significant added value through the productivity channel, and those gains will have profound distributional consequences inside firms and across the economy.
Why is that? It is based on three general ideas that are one part new and one part old but fit within the pantheon of contemporary economics.
- Innovation: First, the ability to produce generalized data and knowledge that results in innovation is nothing new. That ability has been part of capitalism since at least the mid-18th century. If change is the constant, then the integration of AI into the economy and firms will not be as disruptive as conventional wisdom suggests.
- The human factor: Second, AI promises a period in which the technology itself carries an ability to produce original content indistinguishable from content created by human beings. Yet at this point, AI needs to be directed by human beings to do so. Yes, the current vintage of AI can carry out independent functions like teaching itself new ideas and languages to acquire knowledge. But that is still dependent upon human direction, oversight and utilization.
- Democratization: Third, AI will almost certainly result in the democratization and demystification of technical knowledge through natural language processing. This means a lower barrier to entry for organizations, which will result in broad-based increases in productivity, especially around rote tasks that the machines can now do. That democratization will ultimately free up human labor to engage in added-value activity.
One interesting aspect of artificial intelligence is that we can already do much of what it promises. Let’s take a look at the three pillars:
The first pillar of AI is already established, and firms that choose to do so can put it to work.
Much of this pillar has been built on the foundation of mathematics, econometrics and statistical inference.
The revolution over the past 10 years in computing power, cloud storage, big data and predictive analytics provides the foundation for AI to generate text, images and other types of content in a way equal to or superior to humans.
This is not new. For a number of years, businesses have been using predictive analytics to enhance efficiencies, reduce expenses and drive growth.
The second pillar of artificial intelligence is the development around recently produced graphics processing units and large language models. The new capabilities in hardware and software rely upon deep learning to produce what we now call generative AI, which will allow people and firms to increasingly automate non-cognitive tasks, bolstering productivity and innovation. Perhaps more important, that deep analysis of knowledge, text and images now permits AI to create new data and synthetic data that will drive innovation, which future AI models will then use to solve fundamental human and business challenges.
For now, however, large language models use natural language processing to create content and perform rote, or non-cognitive, tasks through human requests.
In addition, they are already demonstrating an uncanny ability to generate answers to human requests, create original text and images, and generate code that can be used for other data science applications.
The third pillar of artificial intelligence is best framed as where we are going. In many respects this is what most people talk about—machine learning and generative AI are a bit too technical to drive a common conversation about the future, but the idea that machines can solve problems is rooted deep in the human mind and experience. For this reason, the third pillar of artificial intelligence can be best described as artificial general intelligence.
That term represents a hypothetical future in which an intelligent agent can derive solutions to observed and non-observed human, scientific and social problems. In such an environment, an autonomous system that surpasses general human intelligence takes over almost all non-cognitive tasks. In doing so, it ensures the optimal use of scarce resources, which in the end enhances social welfare.
We are not there yet, nor will we be anytime soon. But this hypothetical future is on the minds of those who are considering how AI will affect firms, the economy and the human condition—and its manifestation is a goal of artificial intelligence research conducted in both academia and in the corporate sector at firms like OpenAI and DeepMind
How can middle market businesses make the most of a new era in artificial intelligence?
Economic implications: A brief sketch
Enhanced productivity, accelerated innovation and other benefits of artificial intelligence will improve overall social and economic welfare. But those gains will not be equally distributed, especially in the early years. Like the free trade and technology booms of the past four decades, artificial intelligence will have an asymmetrical impact, yielding less benefit for those who resist acquiring the skills needed to use it.
The following is a brief sketch of the economic impact of artificial intelligence.
Productivity and capital spending
Even in the early months of this revolution, there are clear signs that artificial intelligence will generate a productivity boom. These booms are difficult to predict and often have lags of up to 20 years following the initial technological breakthrough. It is reasonable to anticipate an increase in overall productivity of 1.5 percentage points per year in the 10 years following such a breakthrough.
While that scenario is optimistic, a technology boom like the one in the 1990s implies a meaningful increase in productivity-enhancing capital spending over the next five years. That spending will boost growth and productivity, and dampen inflation.
Given that fixed investment in the first quarter of this year was $2.8 trillion, or roughly 10.6% of overall U.S. gross domestic product, the implications of this are significant, as are the distributional consequences for households and firms that do not make such investments.
Households and firms that do nothing or arrive late to the game will lose out on the gains that will follow.
Real yields
An AI-inspired productivity shock clearly has financial market implications. Given the recent elevated inflation and interest rates, with both real and nominal yields establishing multidecade highs, such a shock is a nontrivial concern.
While determining the direction and ultimate level of the real neutral interest rate, known as R*—a non-observable variable—is almost impossible, policymakers still try to do so because the estimation of R* is thought to be consistent with potential GDP growth and price stability.
A productivity shock caused by AI would tend to drive R* higher, as well as both nominal and real interest rates. Given the importance of low interest rates and leverage to non-publicly listed firms that constitute the bulk of the economy, this is a serious matter that may require companies to address their balance sheets and debt ratios.
During the dot-com boom of the 1990s that increased productivity, both R* and real rates moved higher.
Labor market
Early evidence inside the AI ecosystem implies a two to three percentage point boost in labor productivity in the first two to three years. This is a useful starting point for a discussion about what we predict will be an asymmetrical impact on the American labor market.
The first experiments in integrating AI into the workforce have been somewhat surprising. Results suggest that it quickly boosts the productivity and output of cohorts that are not necessarily well positioned to benefit from such sophisticated technology.
A study by Erik Brynjolfsson, Danielle Li and Lindsey Raymond, “Generative AI at Work,” for example, implies that access to the tool increases productivity, as measured by issues resolved per hour, by 14% on average, with the greatest impact on novice and low-skilled workers, and minimal impact on experienced and highly skilled workers.
That study and others suggest a productivity boom in the offing, but it is important to recognize that not all workers will embrace such productivity tools. Not everyone, after all, has the interest or skills to be a cloud computing analyst.
We are confident that this new boom will produce bountiful high-paying jobs. But those gains will not be evenly distributed, and some workers will be displaced and experience a loss of economic status and welfare.
The takeaway
Artificial intelligence is often talked about but rarely understood. At this point one thing is certain: This revolutionary technology will permanently alter the structure of the economy. While we are still in the early days of this revolution, a productivity boom has already begun.
RSM contributors
The middle market AI playbook
Our new guide provides critical AI insights and shows you how to successfully deploy an effective AI strategy.
Topics covered include:
- What AI is and what it isn’t
- Common uses for AI solutions in the middle market
- The AI adoption continuum
- AI readiness
- Taking a responsible approach to generative AI
- Keeping up with evolving solutions
- Getting help on your AI journey
More from The Real Economy
