Building on Your Week 1 Project: Take your highest-priority feature from Week 1 (the one you reframed on Day 6) and identify how it could fail.
Building on Your Week 1 Project: Take your highest-priority feature from Week 1 (the one you reframed on Day 6) and identify how it could fail.
Open your Gemini chat, and add the following:
PERSONA: You are a Chief Risk Officer specialising in AI-native products.
TASK: Produce a failure mode mapping for this feature by identifying at least one way each of the following failure modes could manifest:
- Hallucination: What false information could it generate?
- Bias: Which user groups might receive worse outcomes?
- Drift: What real-world changes would degrade performance?
- Broken handoff: Where could context be lost?
OUTPUT: Output as a table, with an impact assessment rating against each failure mode.
For your highest-impact failure mode, design one guardrail using the pattern:
PERSONA: You are a AI Expereince Designer specialising in ethics.
TASK: Design guardrails for each of the above failure modes:
OUTPUT:
For each one create a short summary using the following format...
- Failure mode: eg... Hallucination (recommending dangerous self-treatment)
- Impact: eg... High (health risk)
- Guardrail: eg... IF confidence score < 85% OR symptoms match high-risk conditions THEN show disclaimer "This is not medical advice" AND require "Speak to Doctor" button click before showing any recommendations
Ask Gemini to design a guardrail for your highest-impact failure mode
💾 Save this Risk Assessment—you'll reference it on Day 8 when designing your data strategy.