I recently attended an intensely engaging conference at the Angelicum in Rome, where I spoke on AI governance and safety.
My contribution focused in a special way on Claude’s Constitution as a unique example of a virtue-based approach to governance and safety.
Despite some critical philosophical questions that such a constitution raises, I argued that the overall approach, including instructing an AI tool to behave in a manner that is good, wise and virtuous, should be considered a promising way of advancing AI safety.
A key takeaway from the discussion was the need to engage not only philosophically and theologically with technology itself, but also with the regulation required to govern it responsibly.