The Economic Implications Of Generative AI: A Sober Analysis

By Brett Hurll

5th September 2023

The advent of generative artificial intelligence (AI), epitomized by OpenAI's large language models (LLMs), has ignited a fervor of speculation and anticipation. While some predict transformative economic growth, others caution against over-exuberance. The question that looms large is: What does the rise of AI mean for the global economy?

The Promise and the Skepticism

Recent studies, such as one by Goldman Sachs, suggest that widespread AI adoption could potentially increase annual global GDP by nearly $7 trillion over a decade. Academic research also indicates that firms adopting AI could see a 3% rise in annual labor-productivity growth. However, financial markets tell a different story. Share prices of AI-focused companies have underperformed the global average in the past year. Long-term interest rates, which would be expected to rise if AI were truly a game-changer, have actually fallen. 

Lessons from History

To gauge the potential economic impact of AI, it's instructive to look at historical technological breakthroughs. The Industrial Revolution, for instance, was not solely the result of a single invention like the spinning jenny but a confluence of factors such as the use of coal, property rights, and scientific advancements. Similarly, Nobel laureate Robert Fogel found that the American railway system, contrary to popular belief, had a modest economic impact, as it merely replaced existing technologies like canals.

Market Dynamics and Monopolies

While AI could potentially generate $430 billion in annual global enterprise-software revenues, this figure is not as transformative as it may seem. Even if all these revenues were converted into profits and concentrated in the United States, the ratio of the country's pre-tax corporate profits to its GDP would only rise from 12% to 14%. Moreover, the likelihood of a single company monopolizing the AI industry appears slim. High costs and proprietary knowledge may create barriers to entry, but the industry is more likely to be dominated by a few large players rather than a single behemoth.

Labor Market Disruptions

The fear that AI will lead to mass unemployment is not new. Previous technological advancements, such as outsourcing and automation, have also been met with dire predictions. However, history shows that job destruction often occurs more slowly than anticipated. Regulatory hurdles, especially in sectors like education and healthcare, can slow technological adoption. Moreover, new types of jobs often emerge as old ones are eliminated.

Productivity Paradox

While AI has the potential to make certain industries more productive, its overall impact on productivity remains uncertain. For instance, sectors like construction and farming, which account for about 20% of rich-world GDP, are unlikely to see significant productivity gains from AI. Additionally, AI could introduce new inefficiencies. For example, the ability of AI to generate large volumes of text could lead to an increase in spam and fraud, requiring additional resources to combat these issues.

Legal and Regulatory Challenges

The proliferation of AI could also result in a surge in litigation, as the costs associated with legal proceedings could plummet. This could lead to an increase in the number of lawyers and further complicate regulatory landscapes.

Conclusion

While AI may bring about changes that are currently unimaginable, it is unlikely to upend the economy in the ways some predict. Just as Fogel's analysis of the American railway system revealed that its impact was less substantial than commonly believed, a future analysis of generative AI may yield similar conclusions. Therefore, while the transformative potential of AI should not be dismissed, neither should it be overstated.