There’s a pattern of AI usage that Ezra Klein mentions that I’ve also noticed.

Since AI tools work best when they have context, there is a push to make all content available to AI or designed for AI first.

This means little small talk in meetings as that ends up in the transcripts that AI produces. Similarly lots of content seems designed for you to consume via AI summaries versus reading it directly as it’s not meant for humans.

We risk optimizing how we work for AI instead of people.