AI Chatbots Will Not Revolutionise The Economy
Which jobs AI changes matters
Traditionally, services have experienced low productivity growth, whereas goods production has experienced high productivity growth. AI can help address this well established economic phenomenon, buoying services productivity growth and triggering major economic expansion.
That sounds like quite an uncontroversial statement; however, it obscures a considerable amount of nuance which really matters. And that is what services actually are.
When people discuss productivity enhancements from AI, at present at least, they are generally talking about the potential for AI chatbots, agents, coding assistants, and the like to improve the day-to-day productivity of services workers. This is typically thought to be workers sat at computers doing repetitive work that AI can help us automate. That might be data entry at a bank, consultants making slides, or automating online chats with customer services.
Separately, we know that the services make up most of the economy in many developed markets. In 2025, just short of 80% of UK gross valued added derived from the services. In Q1 2026, 84% of UK employment was in the services. All that output and all those workers rife for AI automation.
Not quite - the details matter.
Services workers are not one homogeneous mass. Services actually encompass a whole host of activities that AI chatbots haven’t got a chance of automating away. Broadly, services bifurcate into two – a low productivity chunk and a high productivity chunk. For convenience, we split these being either below or above economy wide productivity (as illustrated in the chart below).