Pointing out ineptitudes of current AI is interesting, so far as it goes, but does little to address doomer concerns. It could be seen as a perpetual act of goalpost-shifting, but definitionally, if it isn’t the AI that causes our destruction, it wasn’t the one we were talking about.
Pointing out ineptitudes of current AI is interesting, so far as it goes, but does little to address doomer concerns. It could be seen as a perpetual act of goalpost-shifting, but definitionally, if it isn’t the AI that causes our destruction, it wasn’t the one we were talking about.
No one thinks GPT-4 is that AI. But many were right to think ‘holy shit’.
It’s not too comforting to know human extinction is difficult; starting over with stones as pillows is better than nothing but it doesn’t fill me with as much optimism as does the starship Enterprise.
What’s required is a little imagination. A little David Deutsch. What is possible with intelligence? What is the limit of intelligence?
How alien it will seem. How unknowable.
It’s never too early to start grappling with the implications. After all, which seems closer, a powerful machine intelligence or effective global governance that might mediate something of similar impact?
It’s never a good time to be Cassandra.
Pointing out ineptitudes of current AI is interesting, so far as it goes, but does little to address doomer concerns. It could be seen as a perpetual act of goalpost-shifting, but definitionally, if it isn’t the AI that causes our destruction, it wasn’t the one we were talking about.
No one thinks GPT-4 is that AI. But many were right to think ‘holy shit’.
It’s not too comforting to know human extinction is difficult; starting over with stones as pillows is better than nothing but it doesn’t fill me with as much optimism as does the starship Enterprise.
What’s required is a little imagination. A little David Deutsch. What is possible with intelligence? What is the limit of intelligence?
How alien it will seem. How unknowable.
It’s never too early to start grappling with the implications. After all, which seems closer, a powerful machine intelligence or effective global governance that might mediate something of similar impact?
We have an imagination AND coordination problem.
The problem with the metaphor is that Cassandra was literally clairvoyant. We’re all just bozos speculating about what might happen.
It’s the bozo part that guarantees the outcome.
Only a true bozo — a veritable schmoe, if you will — would assume that any outcome is guaranteed.