Co-Intelligence by Ethan Mollick - some quick thoughts

AI in different roles → he kind of preceded the agentic AI thinking. It's helpful to think of AI agents this way, and to optimize them by having different rules. I think this is going to continue with smaller, more specially-trained models for the particular role/task they have (instead of a “God” model. See post → link to and add commentary about this post.
“Alien Minds” → should it be Alien coding instead of vibe-coding? Alien-assisted coding?
Very practical guide for how you can more broadly think of AI and its role with humanity. I think it's still helpful to think of it as an assistant requiring guidance and feedback, rather than fully autonomous and independent. This guides a lot of my personal philosophy around AI-development and including humans in the loop as the simplest, most effective way of improving quality, security, and reliability, as well as buy-in from your users.
Assum this is the worse AI you'll ever use → so true. It's impossible to stay up with the rate of change. Sidenote: I read something that helped me manage my information intake overload → think of it more as a stream you can dip in and out of, rather than a bucket needing to be emptied. The more recent info will have a recency bias and you'll be more likely to see those, but you can still skim to historical items as you wish.