A Harvard preprint analyzing 103 million app events found fewer switches after AI use, but more switching on days when workers used AI.
The Latest
AI At Work: What The Harvard Study Says About Switching Apps
AI In Hours: A Practical Way For Teams To Assess Its Capabilities
METR’s benchmark puts the best publicly measured AI agent at about three hours of skilled human work with 80% reliability, with important caveats.
What Could APEC Leaders Put On The Table For AI, Trade And Workers?
APEC leaders met in Shenzhen with draft AI and trade proposals before them; adoption was unconfirmed as of Nov. 19, 2026.
OpenAI’s AMD Chip Design Results: What The Up To 14% Figure Covers
OpenAI says its models improved PPA in AMD chip-design blocks by up to 14%. The average, blocks tested and independent verification remain undisclosed.
Anthropic Raises $107.3B: A Look At The Future Of AI
Anthropic’s IPO and private placement total $107.3 billion, but major compute commitments, supplier investments and remaining uncertainties shape the deal.
A PyPI Attack Targeted Microsoft’s AI Agent Framework. Who Takes Action?
Two malicious agent-framework releases were quarantined on PyPI after exposing installed environments to credential theft, researchers say.
Huang Wanted To Return To Nvidia. What Happens Before December 1?
Nvidia says it does not intend to reinstate Jensen Huang as CEO. He is due to testify to House committees on Dec. 1.
Should You Patch Java Now? What CISA’s Java4Shell Alert Means
CISA urges organizations to patch Java4Shell, a Java Base64 decoder flaw. Exposure depends on Java version, processor features and incoming data.
Telephone Operators And AI: How The Job Changed
Automatic switching replaced telephone operators over decades. A historical account traces the job losses, new work and parallels with AI exposure.
Could T3 Code Simplify Working With Claude, Codex, And Several Macs?
A developer says T3 Code brings Claude Code and Codex sessions across eight Macs and a Linux server into one workspace, with beta limits.