Prompted Psyche

Practice

How to Talk to AI

Talking to AI well begins with a clear goal and the context the model needs for the task. State relevant constraints, ask focused questions, and treat the answer as material to examine rather than a verdict. Check important claims and sources, look for uncertainty, and keep decisions with the person responsible for their consequences. There is no single magic prompt. Better results come from guiding the whole exchange: providing context, iterating, verifying, and knowing when not to rely on the output.

Start here

Four exercises for a careful first workflow

Begin with context, then learn how to check sources, expose uncertainty and keep responsibility for the decision.

Working method

It is not only about the prompt

Prompt design can help you ask AI better questions, but responsible work extends beyond the wording of one instruction. It includes a clear goal, relevant context and constraints, suitable sources, iteration, verification, an honest assessment of uncertainty and a person who remains responsible for the result.

The exercises below show how to communicate with AI across that whole process rather than search for a universal formula.

Learning path

Give the model the context it needs

Learn how to give an AI model context by checking whether the goal, constraints, sources and earlier decisions still remain in view.

Learning path

Check the answer

Check an AI answer by tracing its sources, making uncertainty visible and resisting the pull of confident, fluent wording.

Learning path

Use AI to support thinking, not replace it

Use the model as a structured second reader: separate facts from interpretations, test assumptions and ask for counterarguments or alternative readings.

Learning path

Keep responsibility

Organize options, criteria and risks with AI while keeping the final choice with the person accountable for its consequences.