Automation Often Used To Depress Wages, Study Finds
- •Automation since 1980 accounts for 52% of growth in US income inequality.
- •Firms disproportionately use automation to displace high-wage workers rather than boost overall productivity.
- •Inefficient technology targeting has eroded 60-90% of potential automation-driven productivity gains.
The prevailing narrative suggests that automation acts as a rising tide, lifting all boats by boosting industrial productivity. We are often told that machines and algorithms exist primarily to handle the 'dull, dirty, and dangerous' tasks, freeing human capital for more creative, high-value endeavors. However, a sweeping new analysis from the Massachusetts Institute of Technology challenges this optimistic view, suggesting that the reality of the last four decades has been far more calculated and, in many cases, regressive.
Researchers analyzed U.S. labor market trends dating back to 1980, focusing on how firms deploy new technologies. Instead of using artificial intelligence and robotics to maximize overall efficiency, the findings reveal a specific, tactical trend: companies are frequently utilizing automation to dismantle the 'wage premium' of particular worker groups. This practice involves targeting employees who earn slightly above the average wage for their role, effectively using technology as a lever to lower labor costs rather than to drive output.
The economic implications here are profound. According to the research, this specific strategy—targeting high-wage, non-college-educated labor—is responsible for a staggering 52 percent of the growth in U.S. income inequality over the observed period. Rather than boosting productivity, which is the traditional justification for capital investment, this form of automation has paradoxically suppressed national productivity growth.
It is a nuance often missed in the hype cycle surrounding emerging technologies. We often conflate profitability with productivity, assuming that if a firm cuts costs, it must be becoming more efficient. But as the researchers highlight, these concepts are fundamentally distinct. A manager might adopt a new software suite not because it makes the product better or faster, but because it allows for the replacement of expensive, experienced human judgment with a cheaper, automated alternative.
This phenomenon, known as capital-labor substitution, creates a ceiling on growth. When companies focus on rent dissipation—the process of capturing value by suppressing wages rather than creating new wealth—the broader economy stagnates. The study estimates that this inefficient targeting has effectively offset 60 to 90 percent of the productivity gains that should have materialized from automation during the last four decades.
For students looking at the future of work, this is a critical distinction. The next wave of automation will not just be about what machines can do, but about the incentives of the firms deploying them. If the corporate goal is purely cost-cutting, we may continue to see a hollowing out of the middle class rather than a new era of prosperity. Understanding these economic incentives is just as vital as understanding the underlying architectures of AI, because the deployment of technology is ultimately a choice, not an inevitability.