Toward More Efficient Triage With AI & Data
summary
Claims that machine learning and data reduce test cost and delay by prioritizing and quarantining tests, with concrete examples and outcomes: Facebook and others deploy ML test selection to run far fewer tests; an open project carries 18,000 unit tests that take an hour and cost about $10K/month; Dropbox uses quarantine and auto-revert plus reruns to separate flaky infra from real failures; tooling built (Launchable) integrates with CI, uses logs/tracing and probabilistic scoring and reports measurable savings in run time and noise.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 1:49Faster software delivery struggles
- 3:55Triage process in QA
- 7:48Running tests effectively
- 11:05Prioritizing tests with data
- 14:57Multi-stage testing for quality
- 16:20AI test impact analysis
- 18:05Filtering test result noise
- 23:41Quantifying flakiness
- 26:27Snoozing failures
- 27:12Analyzing test failures
- 30:41Monitoring test environments
- 31:25Failure classification
- 34:15Tracking fixes alignment
- 36:40Conclusion
related talks
06mTriple Threat: AI Technologies Reduce Testing Costs
40mFrom Bugs to Brilliance: Leveraging AI in Modern Software Testing
25mWhat’s Our Job When The Machines Do Testing?
01h 04mImpact of AI on #SoftwareTesting: Are Testers Ready? 🔮
41mHow AI is transforming software testing
32mFast, Smart, and Fragile: Rethinking Testing Leadership in the Age of AI
44mHow AI is Exposing the Existential Crisis of Software Testing
25mFuture of Software Testing: Machine Learning Assistance