This is a difficult book to rate. It reminds me of the Black Mirror Episode “Plaything.”
The explanation of LLMs and gradient dissent were nicely done and easy to understand as someone who is a layperson with A.I. The author outlined possible future scenarios for superintelligences. I certainly agree that there will be second order effects in the future that we don’t understand now, and that we should be cautious.
With that said, the scenarios described are very speculative. We as humans are great at forecasting in the short term but not so good at predicting things in the long term. There was a time in the past where researchers thought the world would be overpopulated during the 1960s but as of 2025 most industrialized nations are facing the opposite problem of population decline.
It’s still a great title though.