Skip to content
Slow Variables
Ask
Census · IT & software · occupation code 15-1253

Software Quality Assurance Analysts and Testers

Every task in this role, whether it passes the census's structural hand-over screen, which models pass it, and why. It is a screen, not a claim about what AI can do: a task can pass and still be beyond the systems of the day. The stricter and looser rules re-vote every task, and small differences between roles are noise.

Passes the screen · 4 of 30 tasks

  • Create or maintain databases of known test defects.

    quick to check, an existing check settles it, and a mistake is cheap · 4.7% of the role's time · $1.0B · passed by Claude Sonnet, GPT-6 Sol · the models disagreed

  • Update automated test scripts to ensure currency.

    quick to check, an existing check settles it, and a mistake is cheap · 3.4% of the role's time · $750M · passed by Claude Sonnet, Gemini 3.1 Pro · the models disagreed

  • Modify existing software to correct errors, allow it to adapt to new hardware, or to improve its performance.

    quick to check, an existing check settles it, and a mistake is cheap · 2.8% of the role's time · $620M · passed by Claude Sonnet, Gemini 3.1 Pro · the models disagreed

  • Store, retrieve, and manipulate data for analysis of system capabilities and requirements.

    quick to check, an existing check settles it, and a mistake is cheap · 2.8% of the role's time · $620M · all three pass it

Does not pass

  • Identify, analyze, and document problems with program function, output, online screen, or content.

    reason given most often: no existing check settles it · 8.2% of the role's time · $1.8B · waiting on a check

  • Document software defects, using a bug tracking system, and report defects to software developers.

    reason given most often: no existing check settles it · 7.7% of the role's time · $1.7B · passed by Claude Sonnet · the models disagreed · waiting on a check

  • Install, maintain, or use software testing programs.

    reason given most often: no existing check settles it · 6.0% of the role's time · $1.3B · passed by Gemini 3.1 Pro · the models disagreed · waiting on a check

  • Develop testing programs that address areas such as database impacts, software scenarios, regression testing, negative testing, error or bug retests, or usability.

    reason given most often: no existing check settles it · 5.4% of the role's time · $1.2B

  • Document test procedures to ensure replicability and compliance with standards.

    reason given most often: no existing check settles it · 4.9% of the role's time · $1.1B · waiting on a check

  • Provide feedback and recommendations to developers on software usability and functionality.

    reason given most often: nobody can tell quickly whether it worked, and no existing check settles it · 4.9% of the role's time · $1.1B

  • Monitor program performance to ensure efficient and problem-free operations.

    reason given most often: no existing check settles it · 4.9% of the role's time · $1.1B

  • Design test plans, scenarios, scripts, or procedures.

    reason given most often: no existing check settles it · 4.5% of the role's time · $976M

  • Monitor bug resolution efforts and track successes.

    reason given most often: nobody can tell quickly whether it worked, and no existing check settles it · 4.0% of the role's time · $870M · passed by Claude Sonnet · the models disagreed

  • Test system modifications to prepare for implementation.

    reason given most often: nobody can tell quickly whether it worked, and no existing check settles it · 3.6% of the role's time · $794M

  • Plan test schedules or strategies in accordance with project scope or delivery dates.

    reason given most often: nobody can tell quickly whether it worked, and no existing check settles it · 3.2% of the role's time · $701M

  • Conduct software compatibility tests with programs, hardware, operating systems, or network environments.

    reason given most often: no existing check settles it · 2.8% of the role's time · $620M · passed by Claude Sonnet · the models disagreed

  • Recommend purchase of equipment to control dust, temperature, or humidity in area of system installation.

    reason given most often: nobody can tell quickly whether it worked, and no existing check settles it · 2.8% of the role's time · $620M

  • Identify program deviance from standards, and suggest modifications to ensure compliance.

    reason given most often: no existing check settles it · 2.8% of the role's time · $604M · waiting on a check

  • Perform initial debugging procedures by reviewing configuration files, logs, or code pieces to determine breakdown source.

    reason given most often: no existing check settles it · 2.4% of the role's time · $532M · waiting on a check

  • Visit beta testing sites to evaluate software performance.

    physical work, outside the screen · 2.1% of the role's time · $467M

  • Participate in product design reviews to provide input on functional requirements, product designs, schedules, or potential problems.

    reason given most often: nobody can tell quickly whether it worked, and no existing check settles it · 2.0% of the role's time · $440M

  • Develop or specify standards, methods, or procedures to determine product quality or release readiness.

    reason given most often: nobody can tell quickly whether it worked, and no existing check settles it · 2.0% of the role's time · $434M

  • Investigate customer problems referred by technical support.

    reason given most often: no existing check settles it · 1.9% of the role's time · $422M

  • Coordinate user or third-party testing.

    reason given most often: nobody can tell quickly whether it worked, and no existing check settles it · 1.8% of the role's time · $393M

  • Review software documentation to ensure technical accuracy, compliance, or completeness, or to mitigate risks.

    reason given most often: no existing check settles it · 1.7% of the role's time · $375M · waiting on a check

  • Design or develop automated testing tools.

    reason given most often: nobody can tell quickly whether it worked, and no existing check settles it · 1.6% of the role's time · $348M

  • Conduct historical analyses of test results.

    reason given most often: no existing check settles it · 1.5% of the role's time · $320M · waiting on a check

  • Install and configure recreations of software production environments to allow testing of software performance.

    reason given most often: no existing check settles it · 1.3% of the role's time · $274M

  • Collaborate with field staff or customers to evaluate or diagnose problems and recommend possible solutions.

    reason given most often: nobody can tell quickly whether it worked, and no existing check settles it · 1.2% of the role's time · $269M

  • Evaluate or recommend software for testing or bug tracking.

    reason given most often: nobody can tell quickly whether it worked, and no existing check settles it · 0.8% of the role's time · $174M

Where this role works