Charts, not predictions.

Two ways to navigate uncertain water. The first: build a more sophisticated forecasting model. Hire smarter quants; spend more on data. Predict the wave. The second: read the weather as it changes. Hand back a bearing. Keep the boat heading somewhere honest.

The first one sells better. Models are objects. Forecasts have confidence intervals. People can be measured against their accuracy. The second one is harder to sell because it doesn't produce an artefact; it produces a posture.

Around AI, the first stance has been winning. Vendors pitch models that predict what the workforce will look like, what adoption curves will resolve into, what the next breakthrough will be. We don't know any of these things. The models that pretend to are mostly hindsight reframed as foresight.

We chart, we don't predict. The bearing we hand back doesn't say what will happen. It says: here is where you are. Here is what to try next. We will read the weather again with you next week.