Summary & Key Takeaways ​

• The concept of Directly Responsible Individuals (DRI) originated at Apple, defining the person ultimately accountable for a project's success or failure.
• Simon Willison applies the DRI concept to LLM-powered agents within human organizations.
• He argues that AI agents should never be considered DRIs because accountability is a uniquely human trait.
• An IBM slide from 1979 is cited: "A computer can never be held accountable, therefore a computer must never make a management decision."

Our Commentary ​

This is a crucial discussion. We've been wrestling with the implications of AI agents taking on more complex tasks. The idea that accountability remains firmly human is a grounding thought. It makes me wonder about the legal frameworks that will inevitably need to catch up. There's something unsettling about agents churning away at 3am while nobody's watching, and this piece helps articulate why.