The AI Era (or, as I like to say, AI Crisis) can really be attributed to a single company: OpenAI. As such, it’s an important enough subject that I’m glad someone wrote a book about it before they vanish into a puff of debt.
The folly of OpenAI, thinking they could build a private nonprofit to create aligned AGI when they couldn’t even successfully align and control an Oligarch is an operatic tragedy and the meat of Empire Of AI tells that story. Hao manages to thread the needle between people-zooing the crazies as and the Hagiography of an Isaacson. Genuine appreciation for technological innovation is tempered with empathy for the people the process of its creation is harming. Its perpetrators are painted as human, not cackling villains, though you get the distinct impression of who gave interviews and who didn’t.
Empire also seems to set out to answer a nagging question I’ve had since the beginning of the AI Crisis: how serious are the folks at OpenAI about thinking they’re creating Skynet, and if so why are they doing it? What’s the backstory to the board coup? We get the inside story on both and the answer turns out to be that there were multiple factions within OpenAI (mirroring the opposing factions of accelerationists/safetyists) while still making it clear that both belong to a tiny subculture far out of step with the mainstream which suddenly has massive outsized influence on all of our lives.
A fine sequel to Weapons Of Math Destruction, incidentally.
