AI models have revolutionized the ability of computers to identify useful patterns in massive data sets, but they are incapable of developing causal theories for new scientific discoveries.
Advanced economies are grappling with a prolonged productivity crisis. In the decade following the 2008 financial crisis, hourly productivity growth in the G7 countries collapsed to less than 1% per year, less than half the rate of the previous decade. This poor performance is the biggest economic problem facing developed countries.
Artificial intelligence is a potential game changer. BlackRock CEO Larry Fink says it will transform profit margins across all industries. Goldman Sachs (GS.N) expects productivity growth to increase by up to 3 percentage points per year in the United States over the next decade. The McKinsey Global Institute says it could add up to $26 trillion to global GDP.
Investors need to pay attention to the hype. Four characteristics of AI suggest that while its impact on the bottom line of some companies may be positive, its consequences for the entire economy will be less impressive.
Let’s start with AI’s impact on the most fundamental driver of modern economic growth: the accumulation of new scientific knowledge. Its prodigious predictive powers have enabled notable advances in some data-rich areas of chemistry and biology. However, science’s potential to generate useful knowledge depends not only on its ability to predict what happens but also to explain why it happens.
The ancient Babylonians, for example, were certainly not unprepared when it came to predicting astronomical phenomena. However, they never developed an understanding of the laws of physics that explain why these events occur. It was only with the discovery of the scientific method – building explanatory theories and subjecting them to experimental tests – that scientists began to understand how the universe works. It is that ability to understand as well as predict that allows modern scientists to put a man on the moon.
AI models have revolutionized computers’ ability to identify useful patterns in massive data sets, but they are incapable of developing the causal theories needed for new scientific discoveries. As University of California computer scientist Judea Pearl and co-author Dana Mackenzie say in their 2018 bestseller "The Book of Why": "Data doesn’t understand cause and effect: humans do." Without causal reasoning, AI’s predictive genius will not make human scientists redundant.
A second argument in favor of AI is that it will reduce business costs by automating much of the underlying knowledge work. This is a more plausible claim and there is initial evidence in its favor. A recent study found that the introduction of AI-powered chatbots helped customer support functions resolve 14% more issues per hour. The downside is that the overall impact of these efficiency improvements is surprisingly modest.
Daron Acemoglu of the Massachusetts Institute of Technology estimates that 20% of current job tasks in the United States could be performed by AI, and that in about a quarter of these cases, it would be profitable to replace humans with an algorithm. However, even if this replaced nearly 5% of all work, Acemoglu calculates that overall productivity growth would increase by only about half a percentage point over 10 years. This is barely a third of the ground lost since 2008.
Any revival of economic dynamism would be welcome. The third challenge, however, is that in an important class of cases, AI adoption could reverse productivity gains.
Some of the technology’s initial successes have been in its applications to games. In 2017, for example, Google DeepMind’s AlphaZero program stunned the world by demolishing even its most advanced computer chess rivals. This highlighted the potential of employing AI’s strategic acumen in other competitive contexts such as financial trading or digital marketing. The problem is that in real life – unlike games – others can also invest in AI. The result is that spending that may be rational for a single company is collectively counterproductive.
This implies a fourth characteristic of AI that will deal a more subtle blow to productivity. If a rush to AI makes massive capital investment the minimum requirement just to maintain market share, small companies will inevitably be crushed. Industries will tend towards oligopoly. Competition will tend to decrease. Innovation will suffer and productivity will decline even more.
In 1987, Nobel Prize-winning economist Robert Solow lamented that “you can see the computer age everywhere except in productivity statistics.” The effects of AI may soon be all too evident, but not in the positive way that proponents of the technology expect.
Original article published on Money.it Italy 2024-07-07 07:00:00. Original title: Perché l’intelligenza artificiale non migliorerà la produttività?
