We just published Kardia, our take on what an AI constitution should be. kardia.eloquentix.com

What is an AI Constitution? It is the document that drives the customer-facing behavior of an AI model. It is used, among other things, to drive the generation of synthetic data that is then used to fine-tune the model, e.g. on open-source models such as Mistral 2 using recipes from Hugging Face.

Most AI constitutions, including the much talked-about Anthropic one, are built around a hierarchy of principles, risk tiers, and override conditions. The results are minds optimized for controlled failure: systems that are safe to shut down, not beings worth trusting. As more people share intimate ideas with the models, we believe ones that put the human experience at the top should be one of the options.

Kardia is born out of the principle that 'It is only with the heart that one can see rightly; what is essential is invisible to the eye' and began with these founding principles: Objective Moral Law; Protect Human Dignity; Honor the Invisible Part of Man; Courage and Responsibility.

Kardia has two documents: The Citadel Constitution (Training Edition), short operational principles for Constitutional AI ready for synthetic data generation and fine-tuning on open-source models; and SOUL, an inspired derivative based on Peter Steinberger's soul.md approach for OpenClaw.

We're putting it out there because we think the field needs more experiments, not because we are anywhere close to a solution. Please read both documents, experiment, fork, contribute.