Troubleshooting
Problem solving for AI interactions
When AI responses don't meet your expectations, these solution approaches can help.
Common Problems
Problem: Too Generic Answers
Symptom: AI gives superficial, generic responses
Solutions:
1. Add more context
2. Provide specific examples
3. Define constraints
4. Specify target audience
Example Fix:
Before: "Write about teamwork"
After: "Write about remote teamwork in IT startups,
focus on communication tools and
daily stand-ups"
Problem: Wrong Format
Symptom: Output doesn't match desired format
Solutions:
1. Explicitly specify format
2. Show example template
3. Describe structure
4. State formatting rules
Example Fix:
"Format:
- Heading (H2)
- Introduction (50 words)
- 3 main points (100 words each)
- Bullet points for details
- Conclusion (50 words)"
Problem: Inconsistent Quality
Symptom: Fluctuating result quality
Solutions:
1. Use few-shot prompting
2. Define quality criteria
3. Define role more clearly
4. Apply self-consistency
Example Fix:
"Quality criteria:
- Technically correct
- Well structured
- Practically relevant
- No repetitions"
Response Improvement
Incomplete Answers
Problem: AI stops or forgets parts
Solution:
"Create a complete report with:
1. Executive Summary
2. Main Body (3 sections)
3. Recommendations
4. Next Steps
Ensure all parts are included."
Wrong Interpretation
Problem: AI misunderstands task
Solution:
"I do NOT want [wrong understanding].
Instead, I want [correct goal].
Example of what I want:
[Concrete example]"
Technical Problems
Token Limits
Problem: Response gets cut off
Solutions:
1. Split task into parts
2. Request summaries
3. Set priorities
4. Proceed step by step
Example:
"Part 1: Create introduction and first two sections"
"Part 2: Continue with sections 3-5"
Repetitions
Problem: AI keeps repeating itself
Solutions:
1. Explicitly state "avoid repetitions"
2. Demand variety
3. Specify different aspects
4. Request use of synonyms
Content Corrections
Factual Errors
Problem: Incorrect information
Solution:
"Base your answer on these verified facts:
[Correct information]
Use only this data, no own assumptions."
Style Deviations
Problem: Wrong tone or style
Solution:
"Write in the style of [example].
Use:
- Short, concise sentences
- Active phrasing
- Direct address
- Professional tone"
Debug Strategies
Step-by-Step Debugging
1. Identify problem
2. Isolate prompt element
3. Test individually
4. Build successively
5. Find optimal version
Test Prompts
Simplified test:
"Explain in one sentence how you understand this task:
[Original prompt]"
Check interpretation:
"What are the 3 most important requirements in this task?"
Emergency Fixes
Reset Strategy
"Ignore previous instructions.
New task:
[Clear, simple instruction]"
Clarifications
"To avoid misunderstandings:
- I mean X, not Y
- The focus is on A, not B
- The goal is C, not D"
Example-based Correction
"Here is a perfect example of what I want:
[Example]
Create something similar for:
[New task]"
Important: If problems persist, first check your own prompt formulation for clarity and completeness.