The human premium

Citizens Advice’s September report on digital exclusion calls for people using essential services to be able to speak to a human by phone or in person. That is a proposed safeguard, not a claim that a new universal legal right already exists.

This will surprise nobody who has typed “human” into a chat window six times while a cheerful rectangle suggests an article entitled Have you tried paying the bill you are disputing?

The original promise of automation was that machines would absorb the repetitive work, leaving people free to handle ambiguity, distress and judgement. A number of companies appear to have implemented the opposite model. The machine gets the customer. The remaining human gets a performance dashboard explaining why the interaction took too long.

A chatbot is useful when the problem is common, the answer is stable and escape is easy. It becomes hostile architecture when its real function is to tire people out before they reach somebody authorised to help.

The humane version of AI is not a bot capable of performing empathy. It is a system clever enough to recognise when it should get out of the way.