Practical playbook

Compare difficult options

You have several believable options, priorities that pull against each other, and no single choice that is best for everyone.

Get the criteria, trade-offs and assumptions out into the open, and keep the final call with the person who has to live with it.

15 min
5 steps
Beginner

Use this page in order

Copy the prompt, follow the steps and check the result before using it.

Copy and adapt
Follow the steps
Check the result
Go deeper only if needed

Try it now

Copy this into your AI tool and fill in the brackets.

Remove anything you cannot share first.

Starter instruction

Help me think through this decision without choosing for me: [decision]. Options: [list]. My constraints: [list]. First propose a short set of decision criteria and ask me to correct them. Then compare the options, show the main trade-offs, mark unverified facts and make the strongest case against whichever option currently looks best.

Do it

Tick each step as you go.

The person and AI shape the work at different moments. Your ticks stay in this browser.

0 of 5 done

YouDefine the decision boundaryList the real options, non negotiables, constraints, affected people and what a bad outcome would look like before asking for a comparison.
AIBuild a transparent comparisonPropose criteria, mark assumptions, identify missing evidence and compare each option through the same questions without declaring a winner.
TogetherStress the leading optionChange weights and assumptions, run a pre mortem, invite the strongest counter case and convert unresolved claims into questions or pilot tests.
Your decisionAccept the trade off consciouslyThe accountable person or group chooses the option, states the trade off being accepted and records why the rejected options were not chosen.
You checkTest before committing where possibleVerify factual claims, speak with affected people, obtain professional advice where required and use a bounded trial when uncertainty can be reduced through experience.

See it done

Watch the judgement change the draft.

Before · the honest starting point

Compare two booking systems for a five-person clinic. Non-negotiables: Australian support hours, exportable data and a simple patient experience.

After · the version a person can stand behind

A gated comparison that leaves one support question unresolved on purpose, plus a short pilot to run before any money changes hands.

How it got thereAI first pass, then your correction

AI · a useful first pass
Comparison across support, data portability, patient workflow, setup effort and cost, with System A ahead on setup and System B ahead on export controls.
You · context changes the answer
The clinic owner promotes data portability to a hard gate rather than a scoring line, deletes a support claim that could not be verified, and adds the staff training time nobody had counted.

The decision stays human

Someone must be able to stand behind the result.

The person or authorised group owns the criteria, the weighting, the acceptable trade off and the final commitment. AI can improve the reasoning process but cannot supply organisational legitimacy or bear the outcome.

Verification required

0 of 4 checked

Verify every factual claim, price, legal requirement, capability and implementation assumption that changes the result.
Distinguish hard gates from preferences and do not allow scoring to compensate for a failed non negotiable.
Check who is affected but absent from the comparison and what operational work each option creates.
Use a pilot, reference check or qualified adviser when the uncertainty is testable or the consequence is regulated.

Go deeper

The detail behind each step.

Open only what you need.

How AI and you split the work3 capabilities

Use the capability. Keep the part that needs a person.

AI can make trade offs visible, but a decision is not a table. The person must decide which criteria are real, which are gates, and which consequences they are willing to live with.

Choose a capability for a real job.

Criteria generation and structured comparison

AI can propose decision criteria, define a consistent comparison format and expose missing information before the team begins arguing from preferred answers.

The person contributes: The person removes irrelevant criteria, turns non negotiables into gates, and explains what success and failure mean in the actual operating context.
Do not assume: A balanced looking framework can give equal weight to issues that are not equally important or convert a value judgement into a misleading score.

Scenario and sensitivity analysis

Reasoning and code tools can test how the result changes when costs, demand, timing, staffing or criterion weights move rather than presenting one fixed ranking.

The person contributes: The person chooses plausible ranges, identifies dependencies the tool cannot see, and decides which scenarios deserve attention.
Do not assume: Sensitivity analysis only explores the assumptions entered. It does not discover political, behavioural or implementation risks that were never represented.

Reasons against each option and ways it could fail

AI can make the strongest case against the leading option, imagine failure paths and produce questions for a pilot, supplier meeting or expert review.

The person contributes: The person judges which objections are credible, brings in stakeholders who will experience the change, and turns useful objections into tests.
Do not assume: A persuasive counter case may be rhetorically strong but factually weak, and repeated agreement from the same tool is not independent evidence.

What you bring that AI cannot3 resources

The resources already inside the work.

Values and priorities

You know which outcomes are essential, which trade offs are tolerable and where a numerical score would disguise a genuine principle or obligation.

Implementation reality

You know the staff capacity, technical debt, relationships, change fatigue and hidden work that determine whether a theoretically strong option will succeed.

Stakeholder consequence

You can identify who benefits, who carries the burden, whose voice is missing and which choice will require trust rather than simple compliance.

What goes wrong, and the skill you buildFailure to watch

Failure to watch

The main failure is outsourced judgement disguised as objectivity. A neat matrix can conceal arbitrary weights, missing stakeholders and assumptions that make the preferred option appear inevitable.

Skill strengthened

Structured decision making

Used well, the workflow strengthens your ability to define criteria, expose assumptions, test trade offs, listen to affected people and explain a decision without pretending uncertainty disappeared.

When this is the wrong fit3 signs

Do not use this in place of regulated professional advice on health, law, money or safety.
Do not let a neat comparison table paper over facts that are missing or were never checked.
Be wary if the tool keeps nudging toward the option you already hinted at, and do not treat a second tool agreeing as proof, since both may have learned the same received wisdom.

Tools that can support this3 tools

Evidence backed starting points, not universal winners. Open a guide or ask Bit to help you weigh the fit.

ChatGPT

Its reviewed task guidance covers building criteria, weighing trade-offs and arguing the counter-case for a bounded decision.

Our read, not tested by usSee why it may fit →Claude

Its reviewed guide suits close reading and sustained reasoning over a single decision you can define clearly.

Our read, not tested by usSee why it may fit →Grok

Its reviewed guide includes Multi-agent, which researches a costly purchase from several angles at once and keeps the unknowns and disagreements visible, finishing with questions for the supplier rather than a verdict.

Our read, not tested by usSee why it may fit →

Keep going3 links

Work areaThink through decisions LearnHow to check an AI answer UpdateGPT-5.6 gives ChatGPT more thinking. Most tasks do not need it.

Where this came fromReviewed 19 September 2026

The Human Bit method →Anthropic AI Fluency ↗OpenAI Academy ↗ChatGPT release notes ↗Claude release notes ↗ChatGPT guide, reviewed on The Human Bit →Grok guide, reviewed on The Human Bit →NIST AI RMF guidance on human roles and oversight ↗Microsoft Guidelines for Human AI Interaction ↗

Reviewed: 19 September 2026
Review by: 19 October 2026
Owner: The Human Bit editorial

One quick check