Where Human Value Lives in an AI World

Kind domains have clear rules, consistent feedback, and a closed set of outcomes. Chess, image classification, pattern recognition across structured data. AI wins here because the problem is fully specified and the feedback loop is tight.

Wicked domains have ambiguous goals, context-dependent rules, and consequences that unfold over years. Strategy, organizational design, client relationships, creative direction, diagnosis in novel situations. AI assists here — it doesn't replace, because the problem itself isn't fully specified.

Most real work is wicked. The value humans add in wicked domains is judgment under uncertainty — the ability to act reasonably when the rules don't fully apply. That's not going away.

Codifiable work gets automated. If the work can be written down clearly enough to hand off to a junior employee, it can be written down clearly enough to automate. This used to happen to low-skill jobs. It now happens to white-collar work.

Judgment and context don't get automated — at least not yet. The skill of connecting ideas across domains, reasoning about novel situations, and making decisions with genuinely incomplete information remains scarce.

Unique value lives at intersections. If you can do one thing really well, so can other people, and so can machines. The magic happens at the meeting point of two valuable things — because the combination is rare and hard to copy.

A financial analyst who deeply understands AI systems is worth more than a pure financial analyst or a pure AI engineer. The combination is what's rare. Develop range deliberately.

The right frame isn't 'AI versus me' — it's 'me plus AI versus problems.' The skill isn't using AI; it's knowing when to use it, how to direct it, and how to verify its output.

Gary Kasparov lost to Deep Blue alone. But Kasparov with a computer beat a computer alone. The human in the pair isn't redundant — they're the part that decides what game is worth playing and whether the move makes sense in context.

Management skills — decomposing work, setting context, verifying output, knowing when to intervene — become multipliers in an AI-augmented environment. The people who manage AI agents well will outperform those who simply use them.

This isn't a metaphor. An AI agent running a complex workflow costs 3-10x a simple call and produces outputs that require meaningful judgment to evaluate. The human in that loop isn't a rubber stamp — they're the quality gate.