A consultancy has completed its artificial-intelligence risk assessment by asking the artificial intelligence whether there is anything everyone ought to worry about.
The resulting document describes the technology as transformative, dependable and “unlikely to do the bad thing”. It recommends appointing a responsible human, ideally one whose job description is currently vague enough to absorb the consequences.
The board praised the report’s balance. It contained a reassuring executive summary, a colourful diagram and a section on ethical considerations which everybody agreed was important and nobody could remember afterwards.
An internal reviewer asked how the recommendations had been checked. The project lead explained that a second chatbot had reviewed the first chatbot’s work. When asked who checked the second, he pointed to the meeting running late.
The company has now introduced mandatory human oversight. Employees must click a box confirming they have read each machine-generated output, although the box can be ticked automatically to improve adoption.
Staff remain confident that accountability is clear. If the system succeeds, the programme will be cited as a leadership achievement. If it fails, a junior analyst will be asked why they accepted the recommendation of the tool leadership insisted they use.
