Skip to content
AI Stack Nest
BeginnerLesson 5 of 7 · 2 min read

When Not to Use AI

AI tools are not appropriate for decisions requiring verified current information, professional accountability, or human judgment about specific people.

The most useful thing you can know about an AI tool is when not to use it. Using the tool where it does not belong wastes time at best and causes real harm at worst.

Situations that require a professional

Medical, legal, financial, and psychological situations where the stakes are high and errors have real consequences. The assistant can help you understand concepts and prepare questions. It cannot take professional responsibility for its answer, examine your specific situation, or be held accountable if it is wrong. Use it to get oriented, then consult a qualified person.

Situations that require current, verified information

Current prices, regulations, laws, software versions, news, public health information, company policies. The model has a training cutoff. Its answers on these topics may be outdated or wrong. Use official sources.

Situations that require human judgment about a specific person

Performance reviews, hiring decisions, disciplinary actions, personal references, clinical or educational assessments of individuals. These require knowledge of the specific person and professional judgment. Using AI output directly for these creates risks of bias, inaccuracy, and unfairness.

Situations where transparency matters

Some contexts require you to be clear that a human made a decision: legal documents, official correspondence, work where originality is specifically expected. In these cases, using AI without disclosure may not be appropriate or permitted.

A concrete example

A manager needs to write a performance review for a team member. Using the assistant to structure the review template and check the clarity of their writing is fine. Using the assistant to generate the actual evaluation of the team member's performance, based on a brief description, is not. The evaluation requires knowledge of the person, the work, and the context that the assistant cannot have.

Try it

Review the last five work tasks where you used or considered using an AI assistant. For each one, ask: did the task require professional accountability, current information, or judgment about a specific person? If yes to any of these, was AI the right tool? Use this as a calibration exercise.

Quick check

Why is AI unsuitable for professional advice in medicine, law, or finance?

Quick check

Which situation calls for a human decision rather than AI output?

0 of 7 lessons done in this track