Co-Intelligence

This book might be a good high-level introduction to the topic of LLMs for an unaffected layman, but it mostly read like a watery blogpost to me. Coming in with 200 pages, generous spacing between the lines, and copy-pasted answers from “AI”, it just doesn't offer that much substance. I was hoping for more since this book was recommended to me by a smart person who was, like me, tired of the noise in this space.

The author doesn't have a computer science background which further damaged my trust to the book, but he mentioned studying innovation (in the business context). Also not helpful for the trust were some AI-sounding paragraphs and clunky phrases like “the worse bad outcomes”.

The framing of “I had three sleepless nights asking AI 'deep' questions and omg look what it replied” did not do it for me, although I think I've seen reviews mentioning a conversational tone that worked for them. In my perception, the entire book lacks nuance and constantly hedges bets to paper over that with words like “maybe the opposite is true”. But it doesn't stay consistent with that: after quite some “we can't know”ing, you get sentences like “they will destroy the way we teach before they improve it” in the next breath.

What I did appreciate and what makes this book something other than a quick-and-dirty money-grab on a hype wave is the list of references at the end. This way if the author mentions a study where “the lowest-performing workers became 35% more productive, while experienced workers gained very little” you can check it out for yourself.

I think the smallest chapter at the end with possible future scenarios was the one where the author had the most to say. For the information online he had these possibilities: (1) resurgence of trust in mainstream media, (2) further division into tribes believing what we want, (3) turning away from online sources of news entirely as they stop being useful. I'd rather read a book investigating that.