I came across a piece by Lee Han Chung that offers a rather sobering frame for today’s AI moment: The AI Great Leap Forward

The title is not casual. It points directly to China’s Great Leap Forward of the late 1950s and early 1960s.

(photo generated by ChatGPT)

That effort was ambitious, fast-moving, and highly visible. It was also deeply performative. Local officials, under pressure to show success, inflated production numbers to protect their positions. Reports moved up the chain. Each layer reinforced the story. Signals replaced reality.

The eventual result was system collapse.

The article draws a parallel to current AI work. Not at the level of outcome, but at the level of behavior.

In the AI world, we are seeing systems that appear to reason, appear to understand, and appear to deliver results. And we are also seeing organizational pressure to “AI all the things.” That pressure does not land evenly. It concentrates in the middle, where managers are asked to show progress, demonstrate adoption, and justify investment up the chain.

And so the pattern starts to look familiar.

Dashboards improve. Demos succeed. Metrics move ever upward.

But underneath this performance, the question remains: what actually got better?

Generated output in place of understanding. Automation in place of judgment. Managed signals in place of ground truth.

This is how expectation drifts and systems break down. Not through designed failure, but through perceived success that is mistaken for real progress.

As Mark Twain is often credited with saying: “History does not repeat, but it often rhymes.”

Lee Han Chung is telling us this may be one of those moments.

Worth reading, especially if you are building or reporting on these systems.