Quality engineering
Testing AI systems and software, benchmarking agents, test automation. 5 posts.
AI grades its own homework: why demand for QA is about to come back
Teams ship more AI-written code and understand less of it, and the same model often writes the tests. DORA links AI to delivery instability. QA closes the gap.
Faster delivery needs stronger tests: QA and test automation in 2026
In 2026, 89% of organizations pilot or deploy generative AI in quality engineering, but only 15% have scaled it, while flaky tests are becoming more common.
Can you trust what your AI product says? Testing AI in 2026
AI testing in 2026 means measuring how often your product is wrong: hallucination rates across 26 top models range from 22% to 94% on one benchmark.
What does automation miss? Manual and exploratory testing in 2026
95.9% of top home pages fail detectable WCAG checks, and scanners find only part of the problem. Manual and exploratory testing cover what scripts miss.
70% on one benchmark, 23% on the next: how to benchmark AI agents
AI agent benchmarking in 2026 means testing your own tasks, repeatedly: top models scoring over 70% on SWE-bench Verified solved about 23% on SWE-Bench Pro.