When an organisation establishes an ethics committee, the decision acknowledges that existing people and processes do not meet critical challenges. But there is a paradox here: how to design a fix for a problem you know you don’t yet fully understand.

Last week, Google got caught in that paradox. Having decided that it needed to explore the implications of its work with artificial intelligence, the company attracted fierce criticism for appointing to its new AI ethics committee the head of a rightwing think-tank and the chief executive of a drone company.

Ethics is fundamentally about our responsibility to examine the impact of our decisions on others. Ethics committees, therefore, are accountable to the powerful, but responsible for protecting the powerless. They are routinely used in science and medicine, for example, to oversee work that is experimental, where risks are present but are poorly understood.