Exception Handling
The trade-off between automation and human judgment for edge cases
People are very good at managing exceptions, especially when they are infrequent or easy to resolve. If most initiatives coming from marketing are usable as is and only a small portion need to be combined, renamed, or removed, a human can quickly make those adjustments before a review. The manual effort is tolerable because it is straightforward and occasional.
Machines, by contrast, can process exceptions but usually need explicit structure and instructions. This is why teams often survive with spreadsheets that require some cleanup. The work is a bit messy, but it is manageable. When designing work systems, there is always a trade-off between accounting for every possible exception, which introduces complexity, and keeping things simple while relying on human judgment to handle edge cases.
| The Promise of AI | The Potential Trap |
|---|---|
| AI can expand what is practical to automate by spotting anomalies and suggesting resolutions based on historical patterns. It can support semi-structured exception handling by flagging inconsistencies, proposing merges, or routing decisions, which reduces the manual burden without requiring full automation. | AI can encourage teams to over-automate rare scenarios, adding hidden complexity and maintenance overhead. If people stop exercising judgment, the system may handle the common cases well while mishandling the nuanced ones that actually matter most. |