Why Artificial Intelligence Won’t Boost Productivity

The Impact of Artificial Intelligence on Productivity and Economic Growth

Advanced economies are currently facing a prolonged productivity crisis.
In the decade following the 2008 financial crisis, the growth of hourly productivity in G7 countries has plummeted to less than 1% per year, less than half the rate of the previous decade.
This poor performance is the biggest economic challenge that developed countries are tackling.

Many see artificial intelligence as a potential breakthrough.
BlackRock’s CEO, Larry Fink, claims it will transform profit margins across all sectors.
Goldman Sachs predicts it will boost productivity growth by up to 3 percentage points per year in the United States over the next decade.
The McKinsey Global Institute argues it could add up to $26 trillion to the global GDP.

The Hype and Reality of AI Impact

However, investors need to be cautious about the hype surrounding AI.
Four characteristics of AI suggest that while its impact on individual companies’ bottom lines may be positive, its consequences for the overall economy may be less impressive.

AI’s Impact on Scientific Knowledge Accumulation

Artificial intelligence models have revolutionized computers’ ability to identify useful patterns in vast datasets.
Still, they are incapable of developing the causal theories necessary for new scientific discoveries.
The capacity to understand, not just predict, is what allows modern scientists to achieve groundbreaking advancements.

According to computer scientist Judea Pearl and author Dana Mackenzie, “Data do not understand cause and effect: humans do.” Without causal reasoning, AI’s predictive power will not render human scientists redundant.

Automation of Basic Knowledge Work

One plausible argument for AI is that it will reduce business costs by automating much of the basic knowledge work.
Initial evidence supports this claim, showing improvements in efficiency, such as chatbots powered by AI assisting customer support functions.

However, the overall impact of these efficiency improvements is surprisingly modest.
Despite estimates that AI could handle 20% of current US jobs, with profitability in a quarter of these cases, the overall productivity growth would increase by only around half a percentage point over a decade, according to Daron Acemoglu of the Massachusetts Institute of Technology.

Challenges and Potential Reversals

One significant challenge is that the adoption of AI may reverse productivity gains in certain cases.
While initial successes have been seen in gaming applications, such as Google DeepMind’s AlphaZero defeating top competitors, the real-world implications of AI investments might not yield the expected collective benefits.

If a race for AI dominance makes massive capital investment a minimum requirement for market share maintenance, small businesses will inevitably be squeezed out, industries will trend towards oligopoly, competition will decrease, innovation will suffer, and productivity will further decline.

In conclusion, while the effects of AI on productivity may soon become more evident, they may not align with the positive outcomes that technology advocates anticipate.