🇺🇸 The scariest AI future might be the one where everything works as intended.
I've been thinking a lot about what happens when Artificial Intelligence becomes smarter than the people building it.
So I spoke with Garrison Lovely, author of Obsolete, who has been following some deeply concerning developments inside the AI industry.
Garrison says the immediate danger is already emerging in cybersecurity, where increasingly capable models could allow individuals to launch attacks that once required entire teams of specialists.
But something else from our conversation stuck with me.
We tend to imagine the worst AI scenario as machines breaking free, refusing human instructions and eventually taking control.
Garrison worries about that too. Yet he sees another path to dystopia that doesn't require AI to rebel at all.
What happens if these systems work perfectly, replacing human labor while concentrating enormous wealth and power in the hands of whoever owns them?
Governments could gain unprecedented surveillance capabilities. Autonomous weapons could enforce decisions without soldiers having to obey orders. Ordinary people could lose much of the economic leverage that has historically given them political influence.
And if the machines don't remain under human control, the consequences could be even more profound.
Garrison believes both futures are serious risks, not inevitable outcomes.
I came away wondering whether we're spending too much time asking if AI will obey us, and not nearly enough asking who it will be obeying.
@GarrisonLovely
tweetkeep
