Photo by Pexels

Artificial intelligence is reshaping how software is built, deployed, and verified. Software testing is no exception. From automated test generation and intelligent oracles to flakiness detection and visual regression, AI and machine learning are changing the face of testing in ways that complement—and sometimes challenge—the practices we rely on today. This post looks at where AI is having the biggest impact and how it fits with the automation and testing work we cover on this site.

The rise of AI-assisted testing

Traditional test automation runs scripts that follow predetermined steps and compare outcomes to expected results. It scales repetition and regression, but it still depends on humans to design tests, define oracles, and interpret failures. AI-assisted testing aims to augment that: tools can suggest test cases, infer expected behaviour from existing data, and help prioritise what to run. That does not replace the need for clear requirements and good design—as we argued in The TERMS for Test Automation, tools and technology are one factor among many. But AI is becoming another layer in the stack, especially where large codebases and continuous delivery make manual test design a bottleneck.

Photo by Pexels

Test generation and oracles

Two of the hardest problems in testing are deciding what to test and what the right outcome is. AI can help with both. Machine learning models can analyse application behaviour, user flows, or code structure to propose test scenarios that might otherwise be missed. Some tools generate or refine automation scripts from natural language or recorded sessions. Oracle problem—knowing whether a result is correct—remains difficult; AI can learn “normal” behaviour from logs or past runs and flag anomalies, or suggest assertions from existing tests and documentation. These techniques work best when combined with human judgment and testing heuristics, not as a substitute. As we noted in resuming content, the core challenges of automation are still about people and discipline; AI adds capability but also new questions about bias, coverage, and maintainability.

Flakiness and visual regression

Flaky tests—tests that pass or fail inconsistently—undermine trust in automation. AI can help by identifying patterns in failure data, correlating flakiness with code changes or environment conditions, and suggesting fixes or quarantine rules. Visual regression testing, which checks that UI looks correct, has also been boosted by ML: instead of pixel-perfect comparison, models can learn what “same” and “different” mean in context, reducing false positives. These applications align well with the kind of practical automation and implementation work we discuss in our Automation Chapters.

Photo by Pexels

Accessibility and inclusive design

AI is also touching accessibility testing. Automated accessibility checkers have been around for years; newer tools use ML to better interpret context, suggest fixes, or predict how assistive technologies might behave. That supports the goal of making software usable by the widest possible audience—a theme we care about in our Testing Stories and accessibility testing content. Human evaluation remains essential, but AI can help scale checks and prioritise issues.

What stays the same

AI does not remove the need for skilled testers or clear thinking. Testing is still about risk, coverage, and judgment. Software quality still depends on requirements, design, and collaboration. The resources and heuristics we use—mind maps, challenges, strategy—remain relevant. AI changes the tools and the pace; it does not change the fact that testing is a human-centred, knowledge-intensive activity. Professional development for testers now includes understanding what AI can and cannot do, and how to integrate it into workflows without over-trusting or under-using it.

Photo by Pexels

Looking ahead

AI will continue to change the face of software testing: more intelligent test design, smarter oracles, better handling of flakiness and visual checks, and deeper support for accessibility and other quality dimensions. The best approach is to stay curious, use AI where it clearly helps, and keep the fundamentals—requirements, heuristics, automation discipline, and human judgment—at the centre. For more on test automation foundations, see our blog and Automation Chapters; for community and context-driven thinking, explore And Beyond.