The Harvard Study On Whether AI Can Help With App Switching

A Harvard study of 1,017 workers found more switching on AI-use days, despite less switching immediately after AI use. The findings are observational and under peer review.

How Long Can AI Work Independently? A Guide For Teams

METR’s latest public benchmark puts an AI model’s 80% task horizon at about three hours, with reliability and task messiness limiting its use.

Making Sense Of AI And Work Today — November 19, 2026

A Nov. 19 roundup connects a Rockstar union vote, coding-job data, Anthropic’s funding and two security alerts to questions about work and AI.

The APEC Meeting In Shenzhen: Understanding The AI And Trade Agenda

APEC leaders met in Shenzhen on Nov. 19 with draft AI and digital-trade proposals before them; adoption of the texts was not yet confirmed.

How To Read OpenAI’s Up To 14% Result For AMD Chip Design

OpenAI says its models improved parts of AMD chip designs by up to 14%. The figures are company-reported, with key details and independent checks absent.

What Anthropic’s $107.3B Means For AI Ethics

Anthropic’s record-setting IPO includes $26B from suppliers and rivals. Its prospectus highlights infrastructure costs, governance and safety questions.

Nvidia’s Board Turns Down Huang’s Offer: What Happens Now?

Nvidia’s board says it does not intend to reinstate Jensen Huang. Colette Kress remains interim CEO as congressional interviews approach.

The Microsoft AI Agent Framework PyPI Incident: Who Must Respond?

Two malicious Microsoft Agent Framework releases on PyPI were quarantined on Nov. 17. Users who installed them should rebuild environments and rotate credentials.

Inference Engineering For Software Engineers

Companies are weighing open-model inference providers against frontier APIs to lower costs, tune latency and gain deployment control.

From Human Switchboards To AI Call Routing

A history of U.S. telephone operators shows how dial switching cut jobs for incumbents while younger women moved into other kinds of work.