The AI Boom Hasn’t Made The 15-Hour Week Happen

A 2026 essay compares Keynes’s 15-hour-week idea with income growth, decades of roughly 40-hour workweeks and early AI time-savings evidence.

What Should The UK Consider About AI In Its Youth Employment Plans?

Alan Milburn’s review proposes legal support and AI training for young people who are NEET; the government has accepted its principles but not its benefit reforms.

How European Unions Are Asking Players To Respond To Generative AI In Games

A six-union manifesto urges workers to avoid generative AI, studios to prohibit it and players to support games made without it.

AI’s 2027 Speed-Up: How Long Before Compute Runs Out?

A November 2026 scenario argues AI could accelerate sharply in 2027, but public benchmarks, infrastructure limits and deployment data leave major uncertainties.

GPT-6.2 Astra And Sol: What The New Models Offer At Work

OpenAI released GPT-6.2 Astra and Sol on Nov. 17. Independent tests show strong scores and lower task costs, alongside safety concerns and mixed results.

The Week Nvidia Reports And Anthropic Eyes An IPO: AI’s Possible Takeaways

Nvidia reports on Nov. 17 as Anthropic prepares an IPO. Here are the confirmed timelines, business signals and open questions for AI customers and workers.

Globalization’s AI Losers

Thorsten Meyer’s November 17 report links AI-related trade growth with the uneven job losses seen in earlier trade shocks; labor impacts remain unclear.

Nvidia Rubin HPC And The Prospect Of 50,000-Plus CUDA Cores

Nvidia says its Rubin HPC GPU will exceed 50,000 CUDA cores and target scientific simulations. It is expected in 2028, with key specs still undisclosed.

The Graduate Job Market Is Shifting—But What Can We Attribute To AI?

UK and Australian job data show weaker graduate and entry-level hiring, especially in AI-exposed work, but do not establish AI as the cause.

Claude Fable 5.2 Vs. Gemini 4: Comparing Their Value For Work

Independent tests put Fable 5.2 ahead overall, but Gemini 4 models win several work tasks and cost less per benchmark run.