·
Testing AI-based Systems: Why Traditional QA Fails by Tran Thuy Nha Truc
As AI becomes embedded in every product, traditional QA approaches are no longer enough. AI-based systems behave probabilistically, learn from imperfect data, and degrade over time. This talk explores three core challenges: • Non-determinism (same input, different output) • Bias (models inheriting human and data flaws) • Model drift (performance decay over time) Solution: The AI testing 4R framework (Reliability, Robustness, Responsibility, Resilience)
TAQELAH