Corporate AI adoption is hard: three tough aspects - adoption (getting right tools), infrastructure (compute and data quality), and finding high-value cases where you do something in an AI way end-to-end and throw away the old way. If adoption is less than people think, valuations may be too high.

Impact on jobs is unclear. Some tasks will see greater productivity with fewer people, but also opens capabilities and personalization. Example: sampling all calls for customer service agents improves quality but doesn't reduce headcount. Important to reskill and upskill people.

A job is much more than a collection of tasks.

Seeing creeping impact - things getting marginally more efficient in certain parts. Fundamental impact is coming but not yet visible.