Summary & Key Takeaways ​

• Simon Willison updates a previous analysis of AI adoption rates in large U.S. companies.
• New data from the U.S. Census Bureau's Business Trends and Outlook Survey is used.
• A significant increase in reported AI adoption is observed, from 12% to 40% for large companies.
• This increase is attributed to a change in the survey question from "producing goods or services" to "any business function."
• The article implicitly critiques how measurement methodologies can influence perceived trends in AI adoption.
• The Census Bureau is noted to have fallen foul of Executive Order 14434, related to "Super Intelligence."

Our Commentary ​

This is a fantastic piece of critical analysis. It's so easy to take statistics at face value, especially when they come from official sources. Simon's breakdown of how a simple change in survey wording can completely alter the perception of AI adoption is a crucial reminder to always question the data. It makes me wonder how many other "trends" are similarly influenced by subtle methodological shifts. This is exactly the kind of nuanced perspective we need in the AI discourse.