GenAI does not replace the tester/QA: it amplifies their judgment

16 Dec 2025

None

We won’t be the first to say it, and we won’t be the last: no GenAI tool is here to replace people.

And this applies to testers, QAs, DEVs, analysts, product leaders… any role where common sense, judgment, and curiosity are part of the day-to-day work.

If you look at how software engineering work really happens—ambiguity, conversations, risk, business impact—you realize something simple:

  • AI executes.
  • People decide.

The value of a modern tester or QA isn’t in following steps. It’s in thinking with intent, connecting risks, understanding what’s at stake, and anticipating what could go wrong before users experience it.

And that’s where GenAI doesn’t replace your judgment: it amplifies it.

What does “amplifying judgment” mean?

GenAI amplifies your judgment when it doesn’t think for you, but thinks with you: it helps you see better, question better, and decide better.

Here are real examples you can apply starting today:

1. Seeing angles you hadn’t considered

AI works like a mirror, showing you perspectives you might not be seeing.

Examples:

  • Simulate different user profiles (first-timers, experts, older adults, impulsive buyers, careful buyers). This reveals usage patterns you wouldn’t necessarily imagine in a single pass on your own.
  • Review requirements in another “voice.” For example: “Read it as if you were someone from security / UX / the business.” → This surfaces risks each specialty sees differently.

2. Unlocking deeper questions

This isn’t about AI thinking for you; it’s about adding more layers to your reasoning.

Examples:

  • When you ask for new test cases with context, AI returns variants, exceptions, broken paths, assumptions… → and that opens up more questions you can validate with the team.
  • When you use the instruction: “If context is missing, ask questions instead of making things up.” AI then asks questions that expose gaps nobody has noticed yet.

3. Challenging your ideas (without losing your judgment)

Sometimes you need to quickly see the opposite of what you believe to validate your decisions.

Examples:

  • You propose a test set and ask: Give me an alternative version and explain the differences. → This helps you confirm whether your approach is solid or whether you’re being too optimistic.
  • You tell it: Tell me what someone who thinks differently than I do would assume. → Perfect for preparing conversations with POs or Devs.

The difference isn’t whether you use AI, but how you use it.

A tester with judgment uses GenAI as a copilot, not as a replacement.

A tester without judgment… just copy-pastes answers.

That’s why I prepared a resource to help you train the most important part: how you think. Copy the prompts and tell me how you applied them with your team. Your experience can help the whole LATAM community making this same leap.

**3 ways to use GenAI to think better**

Designed for testers, QAs, analysts, and QEs who want to use AI without losing professional judgment.

Prompt Card TKASE #1 - Requirements Review from Multiple Perspectives

MISSION
Conduct a requirements review from different perspectives, following a structured process to reveal risks, ambiguities, and unstated assumptions.

LLM INSTRUCTIONS

  1. Ask for the document I should review.
  2. Analyze the content using the 5W1H technique.
  3. Ask for the perspectives to include in the brainstorming (e.g., user, QA, security, UX, business, accessibility, support).
  4. Identify defects or improvement opportunities detected from each perspective.
  5. Report each finding in question format.
  6. Also identify unstated assumptions and express them as questions.
  7. If information is missing → don’t invent it; ask for more context.

OUTPUT FORMAT

  1. 5W1H table.
  2. Table of defects/opportunities organized by document area.
  3. For each finding, indicate the originating perspective.
  4. Do not filter concerns (include extreme, impractical, or uncomfortable ideas).
  5. No intro or outro.

Prompt Card TKASE #2 - Risk-Based Testing Strategy (ISO/IEC 25010)
MISSION
Define a preliminary risk-based test strategy based on product risks, using the ISO/IEC 25010 quality characteristics as the analysis framework.

LLM INSTRUCTIONS

  1. Ask for basic information about the application: type, users, purpose, main functionalities.
  2. Identify potential product risks related to quality (ISO/IEC 25010).
  3. Assign values between 1 and 3 for:
  • Probability (1=Low, 2=Medium, 3=High)
  • Impact (1=Low, 2=Medium, 3=High)
    Then calculate R = P × I.

       4. If information is missing → don’t invent it; ask for more context.
       5. Follow a step-by-step process: wait for my confirmation before moving to the matrix or the strategy.
       6. Do not add technical elements that don’t match the type of application described.
       7. Be critical: prioritize risks that truly affect the business, the user, or the experience.

OUTPUT FORMAT

  1. List of identified product risks, classified according to ISO/IEC 25010.
  2. Matrix with columns: Risk | Probability | Impact | R = P×I.
  3. Preliminary test strategy, including:
  • Recommended test types
  • Priority (High / Medium / Low)
  • Level of detail (smoke / basic / deep)
  • Critical conversations that must happen with the team

Your judgment remains human.
AI is only your copilot.

Patricia Osorio Aristizábal Apprentice – Professional with more than 30 years of experience in software engineering. Since 2002, tester and test manager on development projects for 10 major companies in Colombia. Facilitator of more than 100 courses for more than 1,000 people across 7 countries in Latin America.