He had infinite options he failed to exercise, like getting involved in the humanities (sociology, philosophy, law) and/or community building to do the groundwork necessary to contribute positive change rather than illusionary shortcuts to fruitless impact, needless suffering, and self-destruction.
AGI is a canard. Too much money in electricity, servers, land, and technological advances would be needed to build and operate sufficiently capable digital reasoning. Domain-specific pseudo-intelligence task automation will continue here and there with LLMs, deep learning, and such, but no one is building just one "brain" for less than $5T and it probably wouldn't be very performant.
Current AI is totally incapable of replacing expert human judgment for almost everything because they lack constraints, predictability, and reliability. Also, most models hallucinate worthless, incorrect answers so often that they should be relegated to glorified creative, entertainment, and task-acceleration work for the foreseeable future and not be given wildly-inappropriate power like giving a toddler a nuke. Furthermore, the appeal to LLM authority fallacy is a dangerous trend of unthinking, laziness of people failing to think skeptically and find sources for themselves. Finally, too many laypeople and wild-eyed techbros are projecting and imagining AI capabilities that aren't really there and may never exist, but then making much bigger mistakes of irrational, risky bets on their preconceived conclusions, and telling us we should worship the billionaires, accept layoffs, accept panopticon surveillance, accept their loud, polluting data centers in our literal backyards.. for what good?