Microsoft just released its 2026 Work Trend Index, and buried inside the data is a finding that should change how every leader thinks about AI spend for the next 18 months.

The original pitch for AI in business was simple: it would give you time back. Less labor per output, lower cost per output. As recently as the end of 2024, the largest enterprise ROI surveys still showed time-saving as the single most reported benefit of generative AI.

For the last few years, businesses have been living through the free sample phase of artificial intelligence. Powerful models were available for a low monthly fee, teams could experiment with chatbots and coding tools, and employees could use AI for writing, research, summaries, analysis, and workflow support without thinking too hard about the cost behind the scenes.

Most business leaders are still asking the wrong question about AI. They are asking which model they should use. OpenAI, Claude, Gemini, DeepSeek, or whatever open-source model is getting attention this week. That question matters, but it is no longer the most important one.

Thinking back on one of my favorite books, Amusing Ourselves to Death by Neil Postman and how it applies to AI.

DeepMind's new paper, "Dynamic Reflections: Probing Video Representations with Text Alignment," introduces a new approach to understanding video data through the lens of text alignment.

The AI Index Report 2026, produced by the Stanford Institute for Human-Centered AI (HAI), provides a comprehensive look at how artificial intelligence continues to evolve and integrate into various sectors.

There's a widening divide between companies adept at integrating AI and those lagging behind. A new report from McKinsey, the AI Transformation Manifesto, highlights how the frontrunners aren't just dabbling with AI.

Every now and then, you have to step back to move forward.