I'm curious how they reconcile #1 and #2. Could it simply be a matter of adoption in their view? Surely 2023 AI couldn't automate millions of economically valuable jobs.
For me, this is illuminating in two ways:
Everyday, non-software tasks have seen more diminishing returns from recent AI improvements than I expected.
Product (harnesses, terminal tools, NotebookLM-type tools) adoption is lower than I thought.
I'm curious how they reconcile #1 and #2. Could it simply be a matter of adoption in their view? Surely 2023 AI couldn't automate millions of economically valuable jobs.
For me, this is illuminating in two ways: