The Question I Ask Before Every AI Task

I have a single question I ask before I use AI for anything. It is not complicated. But it has changed how I work more than any specific tool or workflow I have adopted. The question is: what is the uniquely human contribution here?

Not "can AI do this?" Almost everything can be partially handled by AI now. Not "should I use AI for this?" That framing puts the tool at the center of the decision. The right question is: in this specific task, what is the part that requires my judgment, my experience, my knowledge of context that no model has access to? And am I protecting that part, or am I outsourcing it along with everything else?

The question forces a useful distinction. Some tasks are almost entirely execution: formatting, drafting from a clear brief, synthesizing information I have already evaluated. For those, AI should do most of the work and I should be reviewing and refining. The human contribution is quality control and judgment about the final output.

Other tasks are almost entirely judgment: deciding what to say and what not to say, reading the political dynamics of a situation, knowing which version of a strategy will actually land with a specific audience. For those, AI might help me think through the problem, but the work is mine. Outsourcing the judgment is not a productivity gain. It is a mistake.

The problem is that these two categories can look similar from the outside. Both involve producing something. Both can be accelerated by AI. The difference is in where the value actually lives, and whether you are protecting it or giving it away.

I have found that asking the question out loud, even just to myself, changes how I approach the task. It makes me more deliberate about which parts of my work I am actually responsible for and which parts I am legitimately delegating. It keeps me from the lazy version of AI use, where you hand everything to the model and accept whatever comes back, which produces outputs that are technically fine and strategically empty.

It also makes me better at prompting. When I know what the uniquely human contribution is, I know what I need to bring to the conversation and what I can ask the AI to handle. The prompt gets more specific, the output gets more useful, and the final product reflects actual thinking rather than the appearance of it.

One question. Before every task. It takes three seconds and it has made every AI-assisted piece of work I produce meaningfully better.