How to Handle Ambiguity?

Someone asked me recently: how do you handle ambiguity?

I paused. Not because I didn't have an answer, but because I wanted to give an honest one rather than a polished one.

Here's what I said: I look at how the attributes of each decision line up to expected value. Then I examine the trade-offs. And for those of us who've been doing this long enough, we have the added advantage of experience - a mental library of situations that rhymes with the current one - plus AI to surface risks and opportunities we might not have thought of ourselves.

That's it. That's my answer.

Expected Value as a North Star

When everything is uncertain, you don't need certainty - you need a framework for reasoning. Expected value is mine. It's not about predicting the future; it's about asking: given what I know, what's the most defensible direction?

It's not perfect. But it's better than gut feel alone, and it's better than paralysis.

The trade-off analysis is the part that slows you down in a good way. It forces you to name what you're giving up, not just what you're gaining. Most bad decisions I've seen - in data, in tech, in business - weren't made because people lacked information. They were made because nobody stopped to ask what are we trading away here?

Experience as a Grounding Mechanism

Here's where seniority actually earns its keep. Not in knowing the right answer, but in having seen enough wrong ones to recognize the shape of a situation.

Ambiguity looks different when you've navigated three platform migrations, two org restructures, and at least one incident that taught you something you'd rather not repeat. Pattern recognition isn't infallible, but it's a real edge - and it's one that can't be easily automated.

Where AI Changes the Equation

This is the part I find genuinely interesting. AI doesn't replace the judgment call - it expands the surface area of what you can consider before making it.

Ask an LLM to steelman the opposing decision. Ask it to list second-order risks. Ask it what you might be missing. Done well, it's like having a tireless devil's advocate who's read more than you have and doesn't have a stake in the outcome.

That combination - experience to ground the decision, AI to stress-test it - is probably the most underrated workflow in the ambiguity toolkit right now.

Is This the Right Answer?

Honestly, I don't know. It's my answer, grounded in how I actually think, not how I think I should think. And maybe that's the point - handling ambiguity well starts with being honest about your own process, not performing someone else's framework.

What's yours?