# Spot where AI could make my team overconfident

Finds the places where polished AI output could lead people to trust, skip checks, or decide too quickly, plus safeguards for each.

- Topic: Working with AI
- Framework: RTF
- Level: Standard
- Tags: Analyze, Fill-in Template
- Source: https://www.thinkwithprompt.com/prompts/spot-where-ai-could-make-my-team-overconfident

## Prompt

```text
Help me find where AI could create false confidence in my team's work, and how to guard against it.

**The work or workflow where we use AI:** [DESCRIBE IT]
**How AI output is used:** [e.g. sent to clients, used in forecasts, informs hiring or pricing decisions]
**Who uses it and their expertise:** [ROLES, experience level]
**Current review practices:** [what gets checked today, by whom]

Look for these overconfidence patterns:
- Fluent, well-formatted output that hides weak evidence.
- Precise-looking numbers, percentages, or forecasts with no traceable source.
- Reviewers rubber-stamping because "the AI already checked it."
- Juniors accepting output in areas where they cannot judge quality.
- Repeated use making people stop reading closely.
- AI agreeing with the user's framing instead of challenging it.
- Gaps: things the AI did not mention that nobody noticed were missing.

Return:
1. A table: Where in the workflow | Overconfidence risk | Warning sign | Likely consequence | Safeguard.
2. The top 3 risks to fix first.
3. Practical safeguards: review questions, required sources, confidence labels, spot-check rates, or second-person review for high-stakes outputs.
4. Three questions every team member should ask before trusting an AI output in this workflow.

Be specific to my workflow. Don't argue for avoiding AI; focus on using it with the right level of trust.
```

## What to fill in

- DESCRIBE IT
- ROLES, experience level
- what gets checked today, by whom

Free to copy and use. From ThinkWithPrompt (https://www.thinkwithprompt.com/), a library of 2,663 curated AI prompts.