Overview
- Major AI firms are actively recruiting philosophers for product and governance roles, sometimes hiring students before graduation and drawing staff away from university departments.
- Some labs have published written 'constitutions' for models, which are sets of rules and principles drawn from sources like Kant, human‑rights documents, and company policies to guide model outputs.
- Commercial products now include user-facing controls that let customers tune model behaviour to different corporate philosophies, an approach exemplified by IBM’s Granite series.
- Teams are applying philosophical methods such as the Socratic method and 'Socratic ignorance' to make models less sycophantic, curb overconfidence, improve chain-of-thought reasoning, and cut hallucinations.
- Experts warn embedding ethics raises unresolved questions about which moral frameworks to adopt, who decides those rules, possible moral deskilling of people, and legal and cultural trade-offs that will reshape careers and product risk.