AI Guides › Career & Work
The Difference Between Looking Busy And Being Useful, In An AI-Assisted Job
By Nigel Guy · 2 min read
AI tools make it easy to produce a lot of visible output fast — more
emails, more documents, more slides, more drafts, all generated in a
fraction of the time they used to take. Producing more visible things feels
like being more useful, and for a while it can even look that way to
everyone around you. It's a different thing from actually being useful, and
the gap between the two tends to surface later than you'd like, usually
right when someone asks what any of it actually changed.
The rule: volume of output and usefulness are different measures, and AI
makes it easier than ever to maximise the one that doesn't matter — the
real test is whether anyone downstream needed less from you because of what
you produced, not how much you produced.
The mechanism: the downstream test
- Pick a recent piece of AI-assisted output you're proud of — a
report, a set of drafts, an analysis — and ask specifically who used it
and for what, not just whether it was well-received.
- Ask whether it replaced work someone else would have had to do, or
simply added to the pile of things now needing review. Output that
creates review work for someone else is a cost, not a contribution, even
when it looks productive from your side.
- Check the ratio of things produced to things actually used or acted
on. A high volume with a low fraction actually used is the classic
shape of looking busy without being useful.
- Ask what you'd cut if you had to justify only the useful half. If
you can name it easily, that's a sign some of your AI-assisted output is
there for appearance rather than function.
- Redirect the freed-up capacity toward fewer, more load-bearing things
— the point of using these tools well is doing less low-value volume,
not filling the same hours with more of it.
What to skip
Skip treating speed of production as the metric worth optimising — a
faster wrong document or a faster generic one isn't progress, it's the
same problem delivered sooner. Skip generating extra drafts, options, or
variations just because it's cheap to do so now; more choices for someone
else to sort through is often a burden dressed up as helpfulness.
Guardrails
- This is a way to check your own output against usefulness, not a claim
about how your organisation actually measures or rewards work — visible
volume is sometimes genuinely what gets recognised, however imperfect
that is, and this framework doesn't change that reality on its own.
- "Useful" still needs someone else's judgement to confirm, not just your
own sense of it — ask directly whether something you produced was
actually used, rather than assuming.
- Being fast and being useful aren't opposites — the goal here is noticing
when speed has quietly become the whole strategy, not slowing down for
its own sake.
All 751 AI guides · JulieMango plans from £17/mo