Just finished “If Anyone Builds It, Everyone Dies” — looking for books on the other side of the argument, and accessible reads on how AI actually works

I've just finished reading If Anyone Builds It, Everyone Dies, which makes the case that developing superintelligent AI poses an existential risk to humanity. I found it compelling but felt there were a lot of assumptions in there, mostly glossed over as axiomatic.

Specifically I'm looking for:

• Books that argue the opposite, that superintelligent AI is either safe, beneficial, or overhyped as a risk
• Accessible, non-technical reads on how modern AI is actually built and trained (LLMs in particular)
• Books that address whether current LLM based approaches could plausibly lead to superintelligence at all, or whether there are fundamental limits

I'm not a researcher, I work in tech but not in AI/ML. I can handle complexity but want readable over academic.