Reporting archive
Evidence on AI and work
Reporting archive

Overnight, Meta swapped headcount for megawatts—gutting people-centric roles, drafting thousands into AI pods, and betting its future on data centers as the real workforce.

The roles standing closest to AI aren’t disappearing—they’re capturing demand and the pay premium.

Security readiness—not cost—just became the choke point for AI value, pushing hiring toward “secure‑AI operators” and rewarding orgs that upskill ops over poaching researchers.

AI didn’t shrink law—it made the cheap parts vanish and the judgment parts decisive, rewiring junior work, pricing, and governance along the way.

Microsoft’s AI chief just turned “12–18 months” from talk into a roadmap—agents as defaults, backfills on ice, and throughput quietly doubling without the pink slips.

Colorado ditches grand AI rituals for a notice‑plus‑liability regime that turns any machine‑shaped HR outcome into evidence if it materially influenced the decision.

AI just jumped to the first line of layoff memos as companies reroute salaries into compute and promise “leverage” long before the metrics catch up.

A G7 central banker just swapped ‘jobs’ for ‘tasks,’ telling us to watch workflows—not layoffs—if we want to see AI’s real impact.

After admitting an AI wrote his speech, Delta’s Ed Bastian tossed it to make a sharper point: in an AI-saturated market, the scarce premium is the person willing to sign the call.

Jensen Huang just told CMU grads that the quickest on-ramp to the AI boom runs through a hard-hat gate, not a Git repo.