An MIT study found that artificial intelligence could worsen income inequality

Findings from the National Bureau of Economic Research indicate that artificial intelligence (AI) may not significantly increase productivity and, on the contrary, may potentially deepen income inequality. Daron Acemoglu, an economics professor at MIT, suggests that predictions about the impact of artificial intelligence on productivity and wages are overly optimistic.

Skepticism about the economic benefits of AI

In his article “Simple macroeconomics of artificial intelligence“Acemoglu examines claims that artificial intelligence will bring significant productivity gains. While some estimates, such as 100 percent GDP growth over the next decade, are highly optimistic, others, such as Goldman Sachs’ projection of seven percent growth in global GDP and 1.5 percentage points of productivity growth over ten years, are more conservative. .

The McKinsey Global Institute report agrees with these lower estimates and suggests annual GDP growth in developed economies of 0.5–3.4 percentage points over the next ten years. Acemoglu remains skeptical, based on historical trends in automation where the benefits have favored business owners and executives while leaving workers with negative consequences.

Moderate performance increase expected

Based on research indicating that approximately 20 percent of U.S. workers could automate half of their tasks using large language models (LLM), Acemoglu projects potential labor cost savings of 27 percent, equating to an overall cost reduction of 14.4 percent. Despite these savings, he says the impact on total factor productivity (TFP) will be limited.

Acemoglu’s calculations suggest TFP growth of just 0.064 percent per year, projecting GDP growth of 0.93 percent to 1.16 percent over the next decade. He cautions that these numbers may still be encouraging as future tasks may be more complex and less amenable to automation. Therefore, it predicts a moderate increase in TFP and GDP, by approximately 0.53 percent, respectively. and 0.90 percent

Concerns about inequality

Acemoglu also discusses the impact of artificial intelligence on wages and economic disparities. He argues that AI is unlikely to significantly raise wages or reduce income inequality. Even if AI increases the productivity of lower- and mid-level workers, it may still have a negative impact on the real earnings of lower-educated women, especially those who are white and native-born. His findings suggest that AI could widen the income gap between capital owners and workers, widening existing disparities.

Last October, leading economists expressed concerns about the divisions artificial intelligence could cause in the workforce. The study, based on government data from a 2018 survey of 474,000 companies (representing approximately four million U.S. businesses), found that less than 6% of companies were using AI technology at the time. However, the adoption rate among larger companies with more than 5,000 employees was over 18%.