Should I use AI for this task?

A better prompt is not always the next step.

Consider AI when you have a clear task, suitable inputs, a way to check the result, and a reason not to use a simpler method. Otherwise, fix the missing condition first. Sometimes the useful decision is to leave AI out.

This guide turns that judgment into six questions and a suggested next step. No AI runs here. The optional check uses fixed rules, not a model or a risk score. It does not inspect your material, check permissions for you, or approve a tool.

Use the first unresolved condition, not a score.

A clear purpose cannot compensate for confidential inputs. Many easy questions cannot cancel one serious consequence. The check therefore uses the following order and still lists every caution that applies.

1. Pause when inputs or consequences are unresolved. If the material is restricted or permission is unknown, do not put it into the proposed tool. If an error could have serious consequences, keep the decision with the appropriate person and arrange qualified review. This guide is not a route to approving high-stakes automation.
2. Prepare before delegating. If you cannot name a useful result or an independent check, arrange those first. For a summary, keep the original notes and compare the important claims. For unfamiliar specialist work, a fluent answer is not a substitute for a reviewer who understands it.
3. Prefer an adequate simpler method. Use an existing checklist, search, calculation or conversation when it already meets the need. If you have not compared methods, try the ordinary approach before calling the AI version a saving.
4. Limit actions, then consider a small trial. Keep sending, spending, deleting and publishing outside the trial. When the other conditions are met, ask for one draft or a few suggestions you can check before using them. Keep the original, define a stop condition, and count review and repair time.

Scope matters: an AI draft can help prepare for a conversation, but it cannot supply another person's agreement. Passing these six questions also does not prove completeness, eliminate bias, or establish that AI will save time. The questions are prompts for your judgment, not a replacement for it.

Similar-looking tasks can need different answers.

Record the boundary before the prompt.

A useful record states what the tool may prepare, what it must not do, how the output will be checked, and when to stop. Keep your completed version in your own notes.

Download the blank AI task decision worksheet (.txt)

SHOULD I USE AI FOR THIS TASK?
Nolan Kido | https://nolankido.com/technology/should-i-use-ai/

A private-use planning worksheet, not a validated assessment or permission to act.
Do not put sensitive information in this blank outline. Keep completed copies private.

Task and useful finished result:

Simpler method to compare:

Inputs I am allowed to use in the chosen tool, and how I checked:

What could go wrong, who could be affected, and how to undo it:

Independent check and person responsible for it:

AI may prepare:

AI may not send, publish, spend, delete, or commit:

My smallest reversible trial, if appropriate:

Stop condition:

What I will count: preparation, production, review, repair and upkeep.

My decision: pause / prepare first / use the simpler method / try a bounded draft.
Reason and date:

Review order: permissions and consequences first; checking and purpose next;
a simpler alternative before a trial. Every unresolved caution still matters.
No result certifies safety, correctness, savings, or approval from another person.

When a bounded trial makes sense

Build the task brief, practice checking a summary, then count the whole workflow's time. These steps answer different questions: what to delegate, whether the output matches its source, and whether the effort was worthwhile.

Prepared with AI assistance. This rule-based decision guide and its examples are authored teaching material, not a validated assessment, a captured AI experiment, or a report of Nolan's personal results.

Broader background: NIST's AI Risk Management Framework addresses managing risks and considering trustworthiness in AI use and evaluation. This small editorial guide is not a NIST assessment, certification, or implementation of that framework.