Reporting archive
Evidence on AI and work
Reporting archive

SXSW’s sharpest takeaway wasn’t a model drop but a mandate: bake worker agency, safety, and job quality into AI’s loss function—or watch automation quietly optimize people out.

Yesterday’s silence on AI and jobs was the tell: the disruption has slipped from headlines into HR portals, budgets, and Monday memos.

DOE just put Genesis on the clock—hands-on AI curricula now and a blueprint supercomputer by 2026—turning a research push into a hiring pipeline.

Atlassian just cut 1,600 roles, split the CTO job in two, and redirected hundreds of millions to an AI substrate aimed at enterprise buyers.

Congress just put a countdown on AI’s labor shock, ordering sector-by-sector forecasts and draft laws on training, UI, and taxes by April 2027.

Beijing is betting it can mint AI-shaped jobs faster than automation erases the old ones—and it’s prepared to stitch the safety net after takeoff.

AI is making white-collar layoffs go quiet while a jammed, underused unemployment system hides the damage and strands workers mid-transition.

Anthropic’s usage data shifts the AI-and-jobs debate from what models could do to which tasks are already running without you—and why junior roles are the first to disappear.

Wall Street’s headcount is shrinking not with pink slips but with empty seats, as AI turns backfills into process reviews.

Oracle is running a blunt experiment: convert payroll into petaflops and see if the market rewards the swap.