The Institute and Faculty of Actuaries has published an AI Manifesto arguing that embedding actuaries early in AI projects, an "actuary-in-the-loop" approach, makes AI systems safer and more reliable, a position that lines up closely with what industry experts have separately identified as the actual technical bottleneck limiting AI's use in insurance pricing today.

One insurance technology specialist said in August that AI is unlikely to be used for pricing in personal lines "anytime soon," because current AI models cannot offer the explainability that traditional rating models provide when insurers need to justify pricing decisions to regulators. That is essentially the gap IFoA's manifesto argues actuaries are trained to close: translating model outputs into something a board or regulator can scrutinise and stand behind.

The manifesto lands as sentiment within the profession has shifted. Actuaries' fear of being displaced by AI fell from 80% in 2024 to 49% in a more recent industry survey, according to pricing platform hyperexponential, alongside a similar drop among underwriters, suggesting growing comfort working alongside AI tools rather than viewing them as a threat.

Real-world examples of the collaboration IFoA is describing are already in market: Aviva has been piloting an AI-powered "Actuarial Agent," built with hyperexponential, designed to support rather than replace underwriters and actuaries in its commercial lines pricing.

Alex Waite, IFoA president-elect and the manifesto's author, said embedding actuaries early means "the system you build is one you can stand behind over the long term."

For brokers, the practical relevance is due diligence: as carriers roll out AI-driven pricing and underwriting tools, whether genuine actuarial oversight sits behind those models is becoming a meaningful question to ask, given IFoA, a body with statutory sign-off authority over insurer reserves and solvency, is now formally positioning that involvement as the difference between AI systems that can withstand regulatory scrutiny and ones that cannot yet.