Ara Kharazian argued that AI's market expansion may increasingly rely on more intensive adoption as the pace of customer additions slows, in Arnold Kling's Oct. 30 roundup.
Kharazian's view was that further market growth may be sustained by more intensive AI use as the increase in new customers loses momentum. The argument puts greater emphasis on adoption intensity as customer additions slow.
Kling's roundup also cited Soumitra Shukla's estimates contrasting AI-feasible work with pressure measured across entire occupations. About 90% of advertised accounting work was AI-feasible, but its pressure score was 8.7 out of 100. That percentage measures advertised work classified as AI-feasible, not the portion of accounting jobs expected to disappear.
Wholesale sales representatives had about 72% AI-feasible exposure, while their pressure score was near 20 because a larger share of their work remained AI-infeasible alongside feasible tasks. Shukla's figures distinguish the share of tasks AI might handle from pressure across an occupation: the sales score was more than twice accounting's despite exposure about 18 percentage points lower.
Isaac Robinson says TaskNet v1 collects experts' descriptions of job tasks through AI-led interviews, according to Kling's roundup. Robinson contrasted this with O*NET's main reliance on surveys, saying interviews can capture forms of work that standard survey questions might not anticipate.
Kling's roundup relayed James Pethokoukis' summary of JPMorgan analysts' visit to the China International Industry Fair, where they assessed humanoid robots against factory automation. In the analysts' view, humanoid demonstrations lagged AI-supported industrial robots in controlled settings on speed and precision, raising questions about near-term returns and repeat orders and leaving the case for wide deployment unproven.
Pethokoukis also cited a separate JPMorgan report about Tesla's Optimus robot. The report identified the hand as its main hardware limit and said Tesla was targeting external commercial shipments in late 2027, with consumer applications expected later.
