Use AI as a sounding board without outsourcing your judgment

Start with the decision, not a preferred answer

A sounding board is useful when you need to organize your thinking, consider alternatives, or rehearse how to explain a choice. It is less useful when you only ask for reasons your favorite idea will succeed.

Write the decision neutrally. Instead of “Help me prove I should launch this course,” try “I am deciding whether to launch a course, run a small workshop first, or delay. Here are my constraints and the evidence I have.”

Separate facts, interpretations, and feelings

All three matter, but they play different roles. A fact might be that three people asked about a topic. An interpretation is that many people will pay for a course. A feeling is excitement about teaching it.

In that fictional example, the requests are encouraging evidence, but they do not establish willingness to pay or the size of the audience. Ask the AI to preserve your enthusiasm while identifying what remains unknown. Skepticism should improve the decision, not punish interest.

Explore more than two options

Decisions often feel harder when framed as all-or-nothing. Ask for a smaller trial, a staged commitment, or a way to gather evidence before committing. Define constraints such as available time, cash, energy, and existing promises.

For a course idea, alternatives might include interviewing potential learners, teaching a short free session, or offering a clearly scoped pilot. Each option has tradeoffs; none should be presented as a guaranteed path to demand.

Invite challenge with a standard

Ask for the strongest case for and against each option using the same criteria. Then ask what evidence would change the assessment. A useful critique identifies a decision-relevant weakness and a way to investigate it.

Avoid endlessly asking for more objections. AI can generate plausible criticism indefinitely, including criticism that does not matter. Set a stopping point: three major uncertainties, one small test, and a date to revisit the choice.

Keep emotional support and decision support distinct

You may want help naming your feelings before evaluating options. Say which kind of conversation you need. “Help me articulate what feels difficult, then ask whether I am ready to compare options” is clearer than asking for a verdict while overwhelmed.

For consequential health, legal, financial, or relationship decisions, use the conversation to prepare questions and organize information. Verify relevant facts and seek qualified or trusted human input where appropriate. The AI has neither complete context nor responsibility for the outcome.

End by writing your own decision and reasoning. If you cannot explain why you chose it without quoting the AI, pause and identify what you still need to understand.

Prompts to put it into practice

Replace the brackets with your context. Use information you have permission to share.

I am deciding between [options]. My goals are [goals], constraints are [constraints], and evidence is [evidence]. Separate facts from assumptions. Compare options using the same criteria, identify three important unknowns, and propose one small next step.

Before recommending a major commitment, suggest a smaller action that could improve this decision. Explain what it would teach us, what it would not establish, and what evidence would justify moving forward.