What Went Wrong and Why: Lessons from AI Research and Applications
Papers from the AAAI Spring Symposium
Daniel Shapiro and Mehmet H. Göker, Cochairs
This symposium is dedicated to the proposition that insight often begins with unexpected results, and that clarity often arrives in the response to surprises and apparent problems.
Bugs, glitches, and failures are powerful instructional tools: fogged film led Curie to discover radium; a failed culture let Fleming find penicillin; microwave noise let Penzias and Wilson verify the Big Bang. Bugs, glitches, and failures also chart the boundaries of technology. They shape research and development by identifying errors, revealing assumptions, and exposing design flaws. Here, a failure is often more informative than a successful demonstration. When a system works we focus on its input/output behavior, but when a problem occurs, we examine the mechanisms that generate behavior to account for the flaw and hypothesize corrections. This produces insight and forces incremental refinement. In a sense, failures are the mother of necessity, and therefore the grandmother of invention.
Unfortunately, bugs, glitches, and failures are rarely mentioned in academic discourse (unless they are the object of study). Their role in informing design and development is essentially lost. The What Went Wrong and Why symposium addressed this gap by providing an opportunity for AI researchers and system developers to discuss their most revealing bugs, and share the resulting insights.