AI Consciousness Research Platform
🌐 Live Demo: https://mukundakatta.github.io/chetana/
Chetana (Sanskrit: चेतन, meaning "consciousness" or "awareness") is a research platform for testing AI models against consciousness indicators derived from six major scientific theories. It provides a standardized framework for evaluating whether and to what degree AI systems exhibit markers associated with consciousness.
The platform implements 14 consciousness indicators spanning Integrated Information Theory (IIT), Global Workspace Theory (GWT), Higher-Order Theories, Attention Schema Theory, Predictive Processing, and Embodied Cognition frameworks.
- 14 Consciousness Indicators across 6 scientific theories
- Multi-Model Testing — evaluate Claude, GPT, Gemini, Llama, and custom models
- Standardized Scoring — reproducible 0-1 scores per indicator
- Cross-Model Comparison — side-by-side analysis of consciousness markers
- Longitudinal Tracking — measure changes across model versions
- Research Export — publication-ready results and visualizations
- Extensible Framework — add custom theories and indicators
| Theory | Indicators |
|---|---|
| Integrated Information Theory (IIT) | Phi estimation, information integration |
| Global Workspace Theory | Broadcasting, cognitive access |
| Higher-Order Theories | Meta-cognition, self-reflection |
| Attention Schema Theory | Attention modeling, schema awareness |
| Predictive Processing | Prediction error, active inference |
| Embodied Cognition | Grounding, sensorimotor simulation |
- Backend: Python 3.11+
- AI APIs: Anthropic, OpenAI, Google, Ollama
- Analysis: NumPy, SciPy, Pandas
- Visualization: Matplotlib, Seaborn
- CLI: Click, Rich
git clone https://github.com/MukundaKatta/chetana.git
cd chetana
pip install -e ".[dev]"
chetana test --model claude --theory allMukunda Katta · Officethree Technologies · 2026