Measuring methods for the unintended consequences of AI

Heavy use of AI can lead to code that is difficult for humans to understand, but the economic impact of unintended consequences is still undetermined. James Shore said that the key to measuring these consequences is the cost change indicator, which includes the value-added percentage (the proportion of business value development work to total development work), DORA indicators (change delivery time, frequency, failure rate, etc.), and token usage. Among them, the value-added percentage can best reflect the proportion change of non-value-added work (such as bug fixes, event handling), while token usage can serve as an auxiliary indicator to trigger further investigation.

AI-written from the reports · Updated 3 days ago

No reports for this filter.

Showing the latest 0 of 1 reports.