Discovering and Validating AI Errors With Crowdsourced Failure Reports

by   Ángel Alexander Cabrera, et al.

AI systems can fail to learn important behaviors, leading to real-world issues like safety concerns and biases. Discovering these systematic failures often requires significant developer attention, from hypothesizing potential edge cases to collecting evidence and validating patterns. To scale and streamline this process, we introduce crowdsourced failure reports, end-user descriptions of how or why a model failed, and show how developers can use them to detect AI errors. We also design and implement Deblinder, a visual analytics system for synthesizing failure reports that developers can use to discover and validate systematic failures. In semi-structured interviews and think-aloud studies with 10 AI practitioners, we explore the affordances of the Deblinder system and the applicability of failure reports in real-world settings. Lastly, we show how collecting additional data from the groups identified by developers can improve model performance.



There are no comments yet.


page 2

page 10

page 16


AI Failures: A Review of Underlying Issues

Instances of Artificial Intelligence (AI) systems failing to deliver con...

Preventing Repeated Real World AI Failures by Cataloging Incidents: The AI Incident Database

Mature industrial sectors (e.g., aviation) collect their real world fail...

Planning for Natural Language Failures with the AI Playbook

Prototyping AI user experiences is challenging due in part to probabilis...

Cutting Away the Confusion From Crowdtesting

Crowdtesting is effective especially when it comes to the feedback on GU...

The Spotlight: A General Method for Discovering Systematic Errors in Deep Learning Models

Supervised learning models often make systematic errors on rare subsets ...

Choreographies meet Communication Failures

Choreographies are global descriptions of communication structures, insp...

Casr-Cluster: Crash Clustering for Linux Applications

Crash report analysis is a necessary step before developers begin fixing...
This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.