The most popular use and the most valuable use are not the same use
Ask a team what they use AI for and you will hear "writing, mostly."
Among US workers who use AI, that is right: writing and editing is the most common use at 51%, with search and research just behind at 49%.
Then ask the same workers where AI actually improves their productivity. The top scores land somewhere else entirely. Coding at 77%. Task automation at 77%. Building presentations at 76%.
There is a plain reason, and it is not skill
Writing is what these tools demonstrate on their front page. It is what the first prompt anybody types produces. No setup, no context, no decision about which part of your week is actually slow.
Your use case was chosen for you, by the demo.
Coding, automation and deck-building have a different shape. Each one starts with a person deciding that a specific, repetitive piece of their work is worth handing over. That decision is the whole difference.
Carry the caveat: these are self-reported, and someone who deliberately picked a task is probably predisposed to like the result. But the three top-rated uses share a property the popular one lacks. Somebody selected them.
You do not find the second use by using AI more
More hours in the writing box will never surface the thing worth automating, because that search does not happen inside the tool. It happens in your own week — looking at what you actually do and naming the piece that is repetitive, rule-bound and quietly expensive.
Which is also why an adoption number cannot tell you much. It goes up either way. It cannot separate a chosen task from a reflex.
The question worth asking your team
Not whether they are using AI. Ask what each person picked it for, and why that one.
If the answer is "writing, mostly," the part of the work that generates the return has not started yet.
The follow-up matters as much: ask what they considered and rejected. Someone who chose writing after weighing two alternatives is in a completely different position from someone who never had the thought. Same answer, different situation — and only one of them needs your help.
Someone in the room while you pick the right task is worth more than another tool: See the workshop →
Keep reading: Find the work with MATCH · Six hours a week correcting AI
This is the thinking behind a Trusted AI Culture.
Sources: Gallup, "Organizational AI Adoption Jumps Six Points" (20 July 2026); Gallup Panel workforce study, 22,573 employed US adults. Use-case figures are scoped to workers using AI for that purpose.