27 September 2026

Welcome to the great AI freakout

Asiatimes | Noah Smith

Americans are increasingly worried about artificial intelligence risk, with a bipartisan majority concerned that the technology is advancing too rapidly and posing potential existential threats. Despite President Trump aligning with an anti-slowdown coalition of investors, libertarians, and China hawks, public opinion and prominent figures like Barack Obama favor safety regulations.

Meanwhile, a public debate persists between AI researchers predicting explosive economic growth rates and skeptical economists who cite structural adoption hurdles and labor barriers. Concurrently, artificial intelligence has failed to disrupt Software as a Service companies, with SaaS revenue and stock prices actually rising as AI acts as a complement rather than a substitute. In other domestic developments, economists Eric Zwick and Owen Zidar highlight the policy impact of privately held pass-through business owners, while conservative commentator Erick Erickson exposes neo-Nazi influence networks attempting to shape the American political right.

Comment

The friction between rapid capability deployment and institutional friction reflects a broader historical pattern familiar from early electrification and the adoption of containerised shipping. When foundational innovations enter commercial markets, productivity gains stall during the necessary reorganisation of enterprise workflows and capital allocation. This transitional lag consistently confounds initial macroeconomic projections of exponential expansion.

Strategic Question for Discussion
Which structural barrier to technology diffusion is most likely to constrain artificial intelligence contributions to productivity over the next decade, and why?
The available evidence points toward the friction of physical automation and enterprise workflow reorganisation as the primary bottleneck. Software and digital tools diffuse rapidly, but restructuring physical production processes requires capital expenditure cycles that inherently outpace software iteration speeds.
Share your assessment in the comments below.