Human Compatible: AI and the Problem of Control Book Price in India, Specifications, Reviews & Offers. Buy online at Amazon.

will edit version in Spanish?
Great video! I've been doing research in AI but I want to understand it more from a philosophical level. Any book recommendations?
Generally I agree, with your assessment of the book (it is a little light-weight given Russell's background). However, I don't think Russell necessarily "strawmans" the opposition. In section 6, "The Not-So-Great AI Debate" Russell presents the arguments of the other side. For some arguments he has a response (e.g. for denial he presents the historical example of nuclear energy), but for many others he seems to describe the problem but not give it any resolution (e.g. for deflection he shows that it is historically possible to control --- in particular ban --- research using the example of cloning, but I doubt he advocates for banning AI research, for tribalism he seems to just say that tribalism is generally an issue). For Russell it is exactly the role of the AI safety researcher (separate from/in conjunction with AI developers) to solve the AI control issues. His primary concern is that current AI developers are not making safety their top priority (which is probably true as many prominent AI groups are still funded by industry). With AGI, it might not be possible to tinker with the program once it starts running due to the potential intelligence explosion so he wants to solve the control issue well BEFORE any AGI ever comes online.
really like the quote from rob reich at 1:04:33
ngl when he started slowing down at 5:12 i thought the machines were taking over