Automation Anxiety: Will A.I. Replace US? The Data Says No

The boom in artificial intelligence is not entirely new to U.S. history. In the 1980s, the
U.S. experienced its “second AI winter.” A strong national interest in competing with Japan in
the global economy and technological innovation gave rise to what were known as expert
systems. These were intelligent computer programs that relied on hardcoded “if-then” rules to
solve problems that typically required human experts. Millions of dollars were poured into these
systems, often overinflating their potential, especially since they could mainly scan and flag
information.
One example of a flaw in expert systems involved a case where a man’s infection was
attributed to an amniocentesis procedure—a procedure performed on pregnant women. The
system appeared to overlook a basic variable: gender. Trust in expert systems was overstated,
particularly as such errors persisted. As these systems failed, jobs were lost, and taxpayers
ultimately bore the cost.
However, overinflated trust in automation continues today, especially with the rise of
modern artificial intelligence. According to MIT economist Daron Acemoglu, AI can perform
only about 5% of human jobs. Significant human oversight is still required for AI to accomplish
even a fraction of what the labor market handles. A Bloomberg Tech article also notes that more
than half of companies regret AI-driven layoffs, particularly given that 75% of S&P 500 growth
has been fueled by AI speculation.
The Remote Labor Index offers a broader view of this issue, representing projects across
the remote economy such as game development, product design, architecture, data analysis, and
video animation. It measures the quality, accuracy, and consistency of both human and
automated work. Findings show that even the best-performing system, Opus, achieved only

3.75% of projected completion with sustained quality. Even fast food jobs have experimented
with replacing workers. McDonald’s, for example, tested an AI-powered drive-through system,
but it frequently mishandled simple orders charging for multiple drinks when only one was
ordered or adding bacon to ice cream.
While artificial intelligence is often framed as a groundbreaking shift, history suggests a
more cautious interpretation. The rise and fall of expert systems in the 1980s should serve as a
recognition that overestimating technological capabilities can harm others. Today’s AI boom
reflects many of the same tendencies such a heavy investment, inflated expectations, and
persistent errors that require human corrections and oversight. Without a more measured
approach, the cycle of overconfidence and costly consequences seen in past technological waves
may continue.

Works Cited

“Expert system.” Wikipedia, Wikimedia Foundation, last modified 16 Apr. 2026,
https://en.wikipedia.org/wiki/Expert_system.

Nano Noodles. “AI in the 1980s and 1990s: The Decades That Changed Everything.” Medium, 4
Nov. 2024,
https://medium.com/@titanmonk90/ai-in-the-1980s-and-1990s-the-decades-that-changed-everyt
hing-a1e8c7a4735a.