Here is my highly-condensed version of the book’s argument, which I find convincing.
We are a long way off from solving what’s known as the alignment problem.
If anyone released artificial superindulgence into the world using anything like our current process, it would almost certainly be misaligned with The interests of humankind.
Artificial super intelligence would have the capability to radically alter, and even extinguish human (and all) life on Earth.
Once it had been released, we would not be able to stop it, contain it, or fix it should it decide to hurt us (or do anything we don’t like).
We are moving towards super intelligence with reckless abandon, rather than an appropriate sense of caution.
Since we don’t know how close we are to achieving super intelligence, the only reasonable way to reduce the existential threat is to globally pause research towards super intelligence.
I sincerely hope this book fuels the global discussions on AI safety.