This is decent as a "rah rah techno-optimism" piece, but it continues Marc's pattern of cheapening the discourse when opining on the topic of AI existential risk.
It's easy to mistake this piece as making an argument for accelerating AI development by touting the virtues of techno-optimism. It's not, though.
The claim that "AI has a high risk of becoming unaligned & uncontrollable in our lifetimes", if sufficiently true (e.g. 99% confidence), implies that the current speed of AI development is unwise, independent of how much one likes techno-optimism in general.
It's easy to mistake this piece as making an argument for accelerating AI development by touting the virtues of techno-optimism. It's not, though.
The claim that "AI has a high risk of becoming unaligned & uncontrollable in our lifetimes", if sufficiently true (e.g. 99% confidence), implies that the current speed of AI development is unwise, independent of how much one likes techno-optimism in general.