More than 50 years ago, Pablo Picasso said computers were useless. They can only give you answers. He wasn’t wrong at the time. Humans framed the problem, computers crunched the solution. A new World Economic Forum piece, “How AI is rewriting the decisions that leaders need to make” argues that boundary has quietly dissolved. AI isn’t just answering anymore. It’s framing. And that is a far more interesting problem than the efficiency story everyone keeps telling.
Frame compression
88% of organisations now use AI in at least one function, up from 55% two years ago. Generative AI usage more than doubled in a single year. All of that has been sold as efficiency. The real shift is upstream of efficiency: AI now gathers the information, spots the pattern, and hands leadership a shortlist of “strategic directions” before the meeting even starts. That sounds like progress. It’s also how you lose the habit of arguing with yourself. When a coherent frame is nearly free to generate, organisations stop paying for the thing that actually made frames good: friction. Debate. The slow, annoying work of asking whether the frame is even right.
Frame convergence
Here’s the part that should worry a boardroom more than compression does. Everyone is now running the same foundation models over their own data. Different inputs, similar interpretive logic, similar blind spots. Errors stop being independent. They start being correlated. An entire sector can be wrong in exactly the same direction at exactly the same time. Keeping “a human in the loop” doesn’t fix this. A human can rigorously stress-test three AI-generated options and never once ask whether those were the right three to consider. That’s not oversight. That’s choosing from a menu someone else wrote.This is, almost word for word, the argument of my own book. Stop Predicting: 13 Lessons from 1,400 Attempts to See What Was Coming.
Start framing
So here’s the question worth asking your own leadership team: how many frames did you actually consider before the AI-generated shortlist showed up? One? Three? Or the number the model happened to give you? Composing the frame, not just choosing from it, is the one job AI still can’t do for you.
