The QA role has been declared “dead” and “more important than ever” in roughly equal measure over the past several years. Neither framing is very useful on its own. This page lays out what the discipline actually covers now, how testing responsibilities are distributed across a modern engineering org, and how AI-generated code is changing the job. For ongoing coverage, see SD Times’ test and testing archives.
Software Testing and QA, Defined
Software testing is the practice of verifying that software behaves as intended — functionally, under load, across edge cases, and over time as the codebase changes. Quality assurance is the broader discipline of building processes that prevent defects rather than just catching them after the fact. In practice, most organizations use the terms loosely and overlapping, but the distinction matters when scoping a role: a “QA engineer” job might be almost entirely manual test-case execution, while a “QA” function at another company is closer to quality engineering — building automation infrastructure and shaping process. The traditional “define expected behavior, run the test, compare results” model is itself under real pressure as AI-generated outputs introduce legitimate run-to-run variation that isn’t a defect in the traditional sense.
The Main Types of Testing in Active Use
- Unit testing — verifying individual functions or components in isolation, typically owned by the developer who wrote the code
- Integration testing — verifying that components work together correctly
- End-to-end (E2E) testing — verifying full user workflows through the actual application
- Regression testing — re-running existing tests after changes to catch unintended breakage
- Performance and load testing — verifying behavior under realistic or peak traffic conditions
- Exploratory/manual testing — human testers probing for issues automated tests aren’t written to catch, especially around usability and unexpected edge cases
Where Testing Responsibility Actually Sits Now
The most consistent structural shift in the last several years is “shift-left” testing — moving testing earlier into the development process rather than treating it as a separate phase after development. In practice this means:
- Developers increasingly write and own unit and integration tests as part of the development task itself, not a handoff to a separate QA team
- Dedicated QA/test engineers focus more on test infrastructure, automation frameworks, E2E test suites, and exploratory testing that’s hard to automate
- CI pipelines run automated test suites on every change, making “testing” a continuous gate rather than a pre-release phase
This doesn’t mean dedicated QA roles have disappeared — it means the role has moved up the stack, from executing individual test cases toward building and maintaining the systems that make automated testing possible at scale.
How AI Is Changing Software Testing
- AI-assisted test generation — tools that draft unit tests or E2E test scripts from existing code, which teams report speeds up first-draft test coverage but still requires human review for correctness and meaningful coverage versus superficial coverage
- More generated code to test, faster — as AI coding assistants increase the rate of code changes, test suites and CI pipelines are under more load, and gaps in test coverage become more consequential more quickly
- Self-healing test automation — tools that automatically adjust E2E tests when UI elements change, reducing one of the most persistent maintenance burdens in E2E suites; this is also explored in the context of broader observability and quality strategy in GenAI-driven development, where intent-driven test generation is positioned as a complement to, not a replacement for, human-set priorities and exploratory testing
- AI-assisted bug triage — using AI to cluster and prioritize incoming bug reports or failing test results, though final triage decisions are still generally human-reviewed
Common Quality Metrics Teams Actually Track
- Defect escape rate — bugs that reach production despite the testing process, often considered the most outcome-relevant single metric
- Test coverage — percentage of code exercised by automated tests, with the caveat that high coverage doesn’t guarantee meaningful coverage
- Mean time to detect / mean time to resolve — how quickly issues are caught and fixed
- Flaky test rate — how often tests fail intermittently for reasons unrelated to actual code defects, a major hidden cost in CI pipeline reliability
Survey data on common quality gaps found that misaligned metrics between technical and non-technical stakeholders, along with weak developer-QA communication, are cited about as often as any specific tooling gap — a reminder that metric selection is itself a cross-functional decision, not just a QA-team one.
Frequently Asked Questions
Is the QA engineer role disappearing as developers write more of their own tests?
The role is changing more than disappearing. As shift-left practices move unit and integration testing onto developers, dedicated QA roles are shifting toward building test automation infrastructure, E2E suites, and handling exploratory testing that’s harder to automate — rather than primarily executing manual test cases.
Does more AI-generated code mean more bugs in production?
It depends heavily on review and testing discipline, not on AI code generation itself. Teams that maintain or strengthen their testing and review processes alongside AI-assisted development report being able to manage the increased rate of change; teams that don’t adjust their process tend to see test coverage and review thoroughness lag behind the increased pace of code changes.
What’s the difference between test coverage and test quality?
Test coverage measures how much code is executed by automated tests; it says nothing about whether those tests would actually catch a real defect. High coverage with weak assertions can look reassuring on a dashboard while leaving meaningful gaps, which is why defect escape rate is generally considered a more reliable quality signal than coverage percentage alone.
What is shift-left testing?
Shift-left testing means moving testing activities earlier in the development process — for example, having developers write and run tests as part of writing the code itself, rather than handing finished code to a separate QA phase after the fact. The goal is catching defects closer to when they’re introduced, when they’re cheaper to fix.
Can AI tools fully automate software testing?
Not currently, for most real-world testing needs. AI tools are widely used to draft test cases, generate E2E scripts, and maintain self-healing UI tests, but human review remains standard practice to confirm that generated tests are actually verifying meaningful behavior rather than just achieving superficial coverage.
