AI & ChatGPT for Beginners

Learn

An AI answer can be biased even without malicious intent. It may reflect incomplete examples, a narrow geographical viewpoint, stereotypes or the wording of the user's question. A leading prompt may encourage the system to favour one outcome. To reduce that risk, ask for alternative interpretations, missing perspectives and evidence for each claim.

Bias checks are particularly important when drafting job advertisements, comparing applicants, explaining unfamiliar cultures or making decisions about people. Avoid using AI alone to decide a person's employment, credit, medical care or eligibility. People affected by consequential decisions should have a fair, accountable review process.

Worked example

Leading: "Why is remote work always worse?" More balanced: "Compare potential benefits and drawbacks of remote and site-based work for a role that includes daily equipment inspections; identify factors that depend on the role." The second framing invites evidence rather than confirmation.

Practice

Rewrite two leading questions as neutral questions. Ask an AI assistant for three reasonable perspectives and identify where the response lacks evidence. Record how the more neutral prompt changes the output.

Key takeaway

Fair prompting requires checking both the question and the answer.