Loading lesson...

Evaluate AI output accuracy — Maîtrise de l'étude avec l'IA

Module 3: Knowing what not to trust — Knowing what NOT to trust is huge. We'll learn how to spot when AI makes things up.

Scenario: You asked AI for 5 benefits of remote work. It listed 5 benefits but didn't mention any downsides or challenges. Which TRACE step catches this?
C — Complete: the answer was one-sided and omitted important information. — The benefits may be real, but leaving out the downsides creates a misleading picture.
Scenario: You asked AI to compare two project management tools. It enthusiastically recommended one but cited a pricing plan that ended last year. Which TRACE step catches this?
A — Accurate: the specific pricing details are outdated and wrong. — The recommendation might be valid, but the specific details (prices) need to be verified.
Scenario: You asked AI to explain quantum computing for a presentation. The response was technically correct but spent 80% of the text on the history of transistors. Which TRACE step flags this?
R — Relevant: the response wandered off-topic from what you actually needed. — The information might be true and accurate, but it doesn't answer the question you asked.
Scenario: You used AI to draft a client proposal and plan to send it straight from the AI output without editing. Which TRACE step should give you pause?
E — Ethical: if the client is paying for your expertise, sending raw AI output without review is dishonest. — This is about your professional responsibility and transparency, not just the content quality.
Scenario: AI tells you that a competitor's product has a specific feature. You try to verify but can't find any mention of that feature anywhere. Which TRACE step flagged the issue?
T — Truthful: the claim couldn't be verified with any independent source. — When you can't verify a claim with any second source, the Truthful check has failed.
After completing the TRACE check, you find that the AI's output passes all 5 steps. What should you do?
Use it confidently — TRACE is designed to give you that confidence when the output checks out. — TRACE is your verification system. When it passes, that's the green light.