Simon Willison argues LLM agents should never be Directly Responsible Individuals

Simon Willison argues that AI agents should never serve as the "Directly Responsible Individual" (DRI) for a project, asserting that accountability is a uniquely human trait. He defines the DRI concept, originating at Apple and documented in the GitLab handbook, as the person ultimately accountable for a specific initiative's success or failure.

Willison contends that machines cannot take accountability for their actions in the way humans can. To support this view, he references an IBM training slide from 1979 which states that because a computer can never be held accountable, it must never make a management decision.