Tim Harford reminds us that innovation is a system, not a gadget. Breakthroughs pay off only when the surrounding parts click into place: printing needed cheap paper and literate readers; shipping containers needed ports, cranes, contracts, and logistics talent; electricity reshaped factories only after workflows, buildings, and training were redesigned.
The implication for AI in insurance is straightforward. Models matter—but complementary investments determine outcomes. Treat adoption as organizational design: redefine roles, incentives, and controls; build skills in data fluency, model governance, and human–AI collaboration; and rework the processes that connect pricing, underwriting, and claims so insights move fast and with accountability. Update how productivity is measured—beyond output—to include cycle time, quality, and customer experience.
Leaders who do this won’t position AI as a replacement for expertise but as an amplifier of it. That is how technology actually raises productivity, strengthens resilience, and earns lasting trust from customers and regulators.
