Human–AI Co-Creativity Consortium (HAI3C)
The Human–AI Co-Creativity Consortium (HAI3C) brings together researchers and practitioners exploring how humans and large language models can co-create scientific insight. Our mission is to build the infrastructure, datasets, and evaluation frameworks needed to train AI systems that think with us — not just for us.
We would like our models to be better scientific collaborators by learning from structured records of scientist–LLM problem-solving sessions.
🧠 Creative Problem Solving in Science
We are building a community of researchers and practitioners who believe human–AI co-creativity can accelerate conceptual breakthroughs, especially in theoretical sciences where progress hinges on ideas, rather than experimental work and data analysis.
Our current focus is to:
- capture interactive trajectories rather than isolated answers,
- model creative exploration, and
- design benchmarks that measure progress toward creative problem solving in scientific discovery.
Learn more: Science →
🚀 Participate
We’re running a closed beta of our guided problem-solving interface.
You can help by contributing your own human–LLM sessions.
- Sign up for the beta: Join the Beta →
- Share use-cases or join discussions: Discussion →
Future Plans
- Public benchmark for creative reasoning and collaborative problem solving
- Arena platform for evaluating LLMs on open-ended creative tasks
- Community data contributions to capture reasoning episodes and “breakthrough points”
Follow development and discussions on our GitHub Discussions.
Ethics & privacy: We publish clear consent terms and protect contributors. See our Privacy page.
Partner Institutions
