Dream-AI, a research project at the International Psychoanalytic University (IPU) Berlin, explores whether modern large language models (LLMs) such as GPT-4o, LLaMA, and Mistral can automate the analysis of dream reports. The project evaluates whether AI can reliably detect psychoanalytic indicators—such as primary and secondary process thinking(Primary Index, PI) and affect regulation and problem-solving capacity (Zurich Dream Process Coding System, ZDPCS)—and whether its assessments match those of trained human coders.
Using the open-access DreamBank database, AI-generated analyses are systematically compared to human ratings. The project is conducted in collaboration with international partners and supported by high-performance computing resources at Charité Berlin. Ultimately, Dream-AI aims to lay the foundation for a scalable AI-based tool to support both research and clinical applications in dream analysis.