Who's responsible when AI gets it wrong? Three questions to answer first
By Rodney Noriega, Founder, Universal Enablement · Published Oct 12, 2026
Short answer
You are. Before any AI workflow goes live, answer three questions. Trust: would your customer be okay with AI doing this? Responsibility: when it's wrong, who on your team owns it? Explaining: can you explain how a result was reached? If any answer is unclear, the workflow isn't ready, however good the tool looks.
In the 1970s, a Stanford system diagnosed infections about as well as some doctors. It was never put into everyday use. Part of the reason was exactly these questions. Would patients trust a machine? If it was wrong, who was responsible? How did it decide?
Fifty years later, self-driving cars face the same questions. So does a 20-person business automating customer replies.
How we answer them in a pilot:
- Trust: start with work customers won't mind being AI-assisted, and tell them where they deal with it directly.
- Responsibility: every workflow has a named person who owns the output. AI drafts, that person approves.
- Explaining: AI's reasoning can't be fully inspected, so we document the inputs, instructions and review steps. You can always explain the process, even when you can't explain the model.
Related
Start with one workflow. In 30 minutes we'll look at where your team's time goes and tell you honestly whether AI can help.
Book a free 30-minute workflow review