The Impracticality of Measuring Near-Magical Gains in Productivity Already Demonstrated by AI

"Through their artificial intelligence model AlphaFold, Demis Hassabis and his team at DeepMind cracked the code. By analyzing the amino acid sequences of hundreds of millions of proteins, the AI learned the rules of protein folding. Today, rather than spending years in a wet lab, the system can predict the highly accurate 3D structure of a protein in just 10 to 20 seconds."

AI completes this incredibly important task that once took a PhD student years in just seconds.

How do you even measure that kind of productivity gain? How would that even show up in productivity gains or economic value?

To the extent that AI is as transformative as it promises to be, it will - like the industrial revolution - actually elude proper measure. Think about air travel. In 1900, no one could travel by air. By 2000, millions of people regularly did. How do you actually measure productivity gains when they go from, "Doesn't exist" to "Fairly common?"