Worried About a Project¶
Something doesn't feel right about your AI project? This journey helps you validate concerns, assess risks, and decide what to do.
Phase 1: Validate Your Concerns¶
Your instincts matter. Use these resources to articulate what's wrong.
Ask the Forbidden Questions¶
Work through questions that probe the areas causing concern:
- About the Model - Is the AI doing what we think it's doing?
- About the Data - Can we trust the data?
- About the Humans - Who's affected and how?
- About Power - Who's really making decisions?
Check for Comfortable Lies¶
Are you hearing reassurances that don't hold up to scrutiny?
Read: Comfortable Lies We Tell Ourselves
Common warning signs: - "The model is just a tool - humans make the decisions" - "We tested it thoroughly" - "The vendor is responsible for that" - "It's the same as what we do manually"
Phase 2: Assess the Risks¶
Now that you've articulated concerns, assess them systematically.
Risk Register¶
Document and score the risks you've identified.
Template: Risk Register
Explore Consequences¶
Think through what happens if things go wrong.
Read: Uncomfortable Futures
Consider: - First-order effects (immediate impact) - Second-order effects (organizational response) - Third-order effects (political and reputational)
Test for Bias¶
If fairness is a concern, run the analysis.
Tool: Bias Detection Tool
Phase 3: Decide What to Do¶
Based on your assessment, choose a path forward.
Option A: Fix It¶
If the project is salvageable, identify what needs to change: - Scope reduction - Additional safeguards - Better governance - Enhanced monitoring
Option B: Kill It¶
Sometimes the right answer is to stop.
Read: When to Kill a Project
Signs it's time to stop: - Core assumptions are wrong - Benefits don't justify risks - Political support has evaporated - Harm outweighs value
Option C: Protect Yourself¶
If you can't change the outcome, protect your position.
Read: Political Survival Guide
Strategies: - Document your concerns in writing - Build allies who share concerns - Create paper trail of warnings - Plan your exit if needed
You're Not Alone¶
Raising concerns about AI projects takes courage. Many projects that failed could have been saved—or stopped earlier—if someone had spoken up.
Resources for support: - Finding Allies - Mapping Resistance - Influence Strategies