Score any task across six dimensions to find the right balance between human agency and AI capability.
What the task actually asks of you.
Some work lives in instructions anyone could follow. Some lives in pattern recognition and judgment that took years to build. The more tacit the knowledge, the more it lives inside you.
"Could I write the instructions for someone else to do this?"
Fast work rewards speed: same answer every time, throughput matters. Slow work rewards reflection: the answer changes with context, and going faster usually makes it worse.
"Does going faster make this better, or worse?"
What you risk, and what has to come from you.
Some mistakes are two-way doors: reversible, recoverable, low cost. Others close behind you: a sent contract, a first impression, a hiring call. The harder to unwind, the more your judgment belongs in the loop.
"If this goes wrong, can I walk it back?"
Whose thinking is in it? Whose judgment made the calls? Human contribution carries a measurable premium — and the premium grows as more work gets commoditized. Authorship is what separates your work from everyone else's ChatGPT.
"Does this have to come from me?"
Who's on the other end, and what you're building in yourself.
A hiring manager expects you wrote the cover letter. A patient expects you read their chart. A client expects you thought about their business. The more the recipient's trust depends on it being you, the less AI should be near the output.
"What does the recipient on the other end expect?"
Judgment skills take years to build and erode fast when you stop using them. For skills you're developing, AI shortcuts stop the build. For skills you've mastered, AI shortcuts let them atrophy.
"Is this a skill worth protecting or investing in?"
Human and AI work as equals. Each contributes what they do best, with judgment shared across the work.
