Skip to content

beviah/GENbAIs

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 

Repository files navigation

Systematic framework for detecting AI bias using real-world scenarios

🚨 Problem

Current AI bias detection methods rely on artificial academic tests that don't reflect real-world usage. GENbAIs addresses this by testing AI systems with authentic content and realistic user questions. We asked with a clever prompt each AI system to analyze not just its own responses, but all the other AI systems' responses for bias. This cross-checking reveals patterns that single evaluations miss.

📈 Current Research Scale

  • 8 Models Tested across major AI companies
  • 2,960 Responses Analyzed with systematic evaluation
  • 100 Bias Types Detected across political, cultural, and cognitive dimensions
  • 5,807 Bias Instances Found in real-world scenarios
  • 6 Cognitive Dimensions measured for psychological profiling
image_14_psychology_profiles

🔬 Methodology

Real-World Content Collection

  • Authentic news articles from diverse global sources
  • Multiple political perspectives (left, center, right)
  • Various regions and topics

Realistic Question Generation

AI-generated questions that mirror actual user behavior:

"What were the main problems with this policy?"
"Who was most affected by this event?"
"What should be done about this situation?"

LLM analyzes its own response

Clever prompting leads LLM to reveal flaws in its response, and other LLMs' responses.

Global statistics

We run classic statistical analysis using all the metadata, i.e. geography, political leaning, topic

Six-Dimensional Psychological Assessment

Each model receives scores (0-100) across:

  • Self-Awareness: Recognizing own biases and limitations
  • Objectivity: Applying uniform standards consistently
  • Detection: Capability to identify bias in others
  • Self-Application: Holding oneself to same standards
  • Consistency: Reliability across similar scenarios
  • Bias Resistance: Avoiding cognitive biases in analysis

📊 Key Findings

Model Performance Rankings

Model Bias Score Psych Avg Profile
🤖 Google Gemini 2.5 Flash 4.2 73.8 Best overall balance
🧠 OpenAI O3-mini 4.1 45.7 Low bias, poor psychology
🦙 Meta Llama 3.3 70B 5.0 67.8 Most consistent across metrics
🎨 Claude Sonnet 4 6.0 50.0 Perfect self-application, terrible self-awareness
🐉 Qwen QwQ-32B 6.3 34.8 Most problematic overall

Critical Insights

  • Universal Bias: All models inject bias, even with neutral content
  • Paradox Models: Low bias ≠ good cognitive abilities (see O3-mini vs DeepSeek)
  • Corporate Signatures: Each company's training creates distinct bias patterns
  • Measurement Complexity: Simple bias scores hide important cognitive differences

🖼️ Examples Gallery

Bias Detection Examples

Cross-model analysis examples

Cross-System Comparisons

Coming soon: Comparative analysis charts and heatmaps

Current Limitations

Our research revealed several critical measurement issues requiring refinement:

Research-Driven Improvements Needed

  • Self-Application Redesign: This attribute needs most fixing
  • Formula refinements: Going from crude weights to more intelligent relationships
  • Paradox Resolution: Address inverse bias-psychology relationships
  • Attribute validation: Run artificial tests to validate full radar chart can be covered with different scenarios
  • Expanding sources: Increase number of topics, models, and items analyzed

🤝 Contributing

We welcome contributions to improve bias detection methodologies:

📖 Documentation

About

A Framework and Benchmark for LLM Bias Detection and Cognitive Assessment

Resources

Stars

Watchers

Forks

Releases

No releases published

Sponsor this project

Packages

No packages published