🧐 A Look at Possibilities of Using AI in #SoftwareTesting
summary
Claims that AI can handle specific testing tasks not whole testing workflows and demonstrates using ChatGPT to generate test data in 10 languages with a language/challenge table and one example per language, plus recommends mimesis/python for synthetic data. Argues LLMs suit interpretive, fuzzy problems like identifying duplicate tests but not trivial deterministic math tasks such as boundary value analysis, warns about hallucination, stresses cost and feasibility tradeoffs, and recommends composing LLMs with other tools and human validation.
AI-written summary · about
chapters (5)
- 0:00Introduction and approach
- 2:23Analyzing new technology
- 10:32Applying framework to AI
- 17:51Practical experiment with LLMs
- 27:42Q&A and closing thoughts
related talks
32mMaryia Tuleika: Real-Life Guide to AI in Software Testing
01h 04mWhat are we thinking in the age of AI?
44mAI Empowered Software Testing
40mShould We Let AI Take Over Test Automation Completely?
44mNatalia Petrovskaia: The AI Illusion: Rethinking the Role of Human Expertise in Software Testing
40mFrom Bugs to Brilliance: Leveraging AI in Modern Software Testing
55mAI-First Testing
25mAI vs. AI: How to Test AI Systems Using AI