I very recently participated in a two-day session to offer guidance on rewriting the constitution for Claude.  The small group invited was uniformly excellent, we received serious time with key decision-makers, and the discussions were of very high quality.

Some of the points I stressed were the following:

1. Whatever one might take a “constitution” to mean in this context, it needs to borrow more from analogs to case law and the common law.

2. Along related lines, think more in terms of “Talmud,” and not just in terms of “Torah.”

3. Work to help build out a quality secondary literature on the AI constitutions and related documents.  Currently this does not exist.

4. Consider how a panel of diverse AIs, with different prompts, could help to evaluate to what extent Claude (and other AI models) were acting in accord with their constitutions.

5. Have a final board of human adjudicators, functioning in a manner analogous to an independent judiciary.  To the extent the panel of diverse AIs might have concerns about Claude not following its constitution, those AIs could alert the human adjudicators to what was going on.  Those human adjudicators could then have authority over potential changes and remedies.

Here a recent short post on using internal courts and the common law to help govern/self-govern AI.  And on the courts.

I thank Anthropic for having us in.