I worry that the usage data on which AI exposure indices are built is sufficiently removed from labor market disruption, and what these indices actually are capturing sufficiently ambiguous to 90%+ of people, that they are actively harmful for getting policy right. I wrote something about it.
Nobody knows what ‘AI exposure’ means
Link post
I worry that the usage data on which AI exposure indices are built is sufficiently removed from labor market disruption, and what these indices actually are capturing sufficiently ambiguous to 90%+ of people, that they are actively harmful for getting policy right. I wrote something about it.