وقتی دانش ارزان می‌شود، چه چیزی ارزشمند می‌ماند؟

This has happened before. When information became cheap to distribute, first through print, later through the internet, the value of simply possessing information fell, and the value of being able to judge, curate, and act on it rose in its place. AI is running a version of this same pattern on a much larger category: not just information, but analysis, synthesis, and increasingly, well-structured reasoning itself.

The pattern, stated plainly

When a category of knowledge becomes cheap, the economic value doesn't disappear from the system. It relocates, usually toward whatever remains scarce once the newly cheap thing is available to everyone at once. After the printing press, the scarce resource stopped being access to a text and became the ability to interpret and use it well. After the internet, the scarce resource stopped being access to information and became the ability to filter and evaluate it.

What's becoming cheap now

AI is making a specific, valuable category of cognitive work genuinely cheap: structured analysis, first drafts, summarization, and increasingly, reasoning that follows an explicit chain of steps. This is a larger and more central category than information distribution was, which is part of why the disruption feels less contained.

What is likely to become the new scarce resource

Following the pattern, the scarce resource should shift toward what AI is still comparatively weak at, and what remains genuinely hard even once analysis itself is cheap: knowing which question is worth asking in a specific, messy, real-world context; holding organizational and political nuance that never gets fully written down anywhere; and being willing to take responsibility for a decision when the analysis, however good, runs out and a call still has to be made.

None of those three are things AI clearly does today. They may become easier over time, but they are structurally different problems from producing good analysis, because they depend on context, stakes, and accountability that live outside any document a model can be shown.

Where this leaves me

If the pattern holds, the practical response isn't to compete with AI on the newly cheap thing, producing more analysis, faster, but to invest deliberately in the three things above: framing better questions, holding context that isn't written down anywhere, and being willing to own a decision. Those look, from here, like the closest thing to durable value in a world where the analysis itself is no longer scarce.