COOKED
The modal QA analyst writes test cases from requirements, builds and maintains Selenium/Cypress/Playwright scripts, runs regression suites, triages failures, and files structured bug reports — all text-and-screen work that coding models now do at usable quality and at volume. What resists is the thin senior tier: exploratory testing that finds the bug nobody specified, owning release go/no-go, designing test strategy across flaky distributed systems, and performance/security testing on real devices and hardware. There is no license, no signature requirement, and buyers do not pay for a relationship with a tester, so nothing outside the work itself slows displacement.
Roughly flat across the period, with year-to-year wobble.
This line is counted by the Bureau of Labor Statistics — the one figure on this page that isn't a judgement of ours. Headcount moves on demand, offshoring, demographics and the business cycle, and automation is one term among several, often not the loudest.
So a falling line is not evidence that AI did it, and a rising one is not evidence that it won't. Both happen in this register: some occupations resist automation and shrink anyway, others are highly automatable and keep growing.
BLS projection, 2024–2034
+10% 201,700 → 221,900 on the projections basis
Exposed, but growing
AI can already do a lot of these tasks, and the BLS still expects +10% more of these jobs by 2034. Demand for the output is growing faster than the work is being automated away — the mechanism BLS gives for software developers, and the combination people most often misread as an error.
Different clocks. The score is what current AI could do to this work today. The projection is how many of these jobs will exist in 2034. Everything between the two — how fast employers actually adopt, whether demand grows in the meantime — is why they can point opposite ways without either being wrong.
~14,000 openings a year on average, including replacing people who leave.
Beta TesterGame TesterData ModelerTest EngineerSystems TesterMalware AnalystQuality AnalystServer EngineerSoftware TesterSystems AnalystQuality EngineerAutomation TesterBug Bounty HunterPerformance TesterSoftware InstallerSolution ArchitectTechnology AnalystUsability EngineerApplication AnalystApplications TesterComputer ConsultantSystems CoordinatorValidation EngineerAutomation Specialist
Holding it up: judgment & accountability . Weakest point: liability shield .
Mixed — a routine tier and a judgment tier Writing test cases from a Jira ticket, generating Playwright selectors, and triaging a red CI build are exactly the text-in/text-out loops LLMs handle now, which caps this at 9 rather than mid-teens; the 9 instead of 4 comes from work that still needs a human in the loop — reproducing a heisenbug that only appears on a specific Android build, deciding which of 300 failing assertions are real versus a selector drift after a UI refactor, and exploratory sessions where the oracle is your own judgment about what the product should do, not a written requirement.
Fully desk- and screen-based A 4 reflects that the job is a laptop, an IDE, and a browser for most sprints, with the physical component limited to the device lab — plugging a phone into a USB hub for real-device testing, checking a kiosk or POS terminal build, or verifying a hardware peripheral integration — none of which happens in an uncontrolled environment or on most teams' sprints.
No licence, no signature requirement There is no state license, no ISTQB requirement in any statute, and no signature on a release; when a defect reaches production the postmortem lands on engineering management and the release owner, and QA's sign-off is an internal Jira transition with no legal weight, which is why this sits at 2 and not 5.
Meaningful discretion Most days are bounded by an acceptance criteria list and a defined severity/priority matrix, which holds this under the discretion band; the 10 comes from calls that genuinely are yours — assigning severity on an ambiguous edge case, deciding a flaky test is masking a race condition rather than muting it, and recommending go/no-go into a release meeting where you can be overruled and usually are.
The verdict above describes this occupation as a whole. Almost nobody does the typical version of a job — tick what's actually in your week and see how your own mix sits.
Your task mix speaks to task resistance (9/20 here) — how much of the day's work current AI already does. That is the dimension the boxes above are about.
It cannot move the other three. Liability shield (2/20) is whether the law requires a licensed human to sign. Trust premium (5/20) is whether buyers specifically pay for a person. Judgment and accountability (10/20) is whether the role exists to own consequential calls. Those are facts about the occupation's standing, not about which tasks are in your week — a paralegal who does only trial exhibits still holds no licence. Together they are 17 of this occupation's 30 points (57%).
Embodiment (4/20) is also a property of the work rather than the worker, but we don't tag individual tasks as physical or not, so the picker can't tell you anything about it. That's a limit of this tool, not a claim.
Did we get the list right? Tell us what's missing — the tasks are written from the outside, and you're reading this from the inside.
Software Developers EXPOSED
Computer Hardware Engineers EXPOSED
The moves above are yours to make. This is the other half: what would have to change in the world for the occupation itself to score higher. None of it is in any one person's gift, but it is where the floor actually comes from. Scores here are not a one-way ratchet. Only two of the five dimensions — task resistance and embodiment — track what machines can do. The other three track law, what buyers will pay for, and who is answerable, and those move in both directions, often in response to the same pressure AI creates. If every lever below landed, this occupation would score around 46/100 — EXPOSED.
Task-mix shift: once script authoring and regression triage are model-generated, the surviving role is exploratory/adversarial testing, oracle design for nondeterministic distributed systems, and validating AI-generated tests against real requirements. Watchable signal: job postings retitled 'test architect' or 'quality engineer' with script-maintenance duties dropped and 'AI test review' added. Raises the residual, but shrinks headcount.
Safety-critical software regimes naming a human verification role: FDA cybersecurity/premarket software guidance and IEC 62304, DO-178C DAL-A independent verification, and EU AI Act Art. 17 quality-management/post-market monitoring for high-risk systems already require documented, traceable verification with named responsible personnel. If auditors or notified bodies begin rejecting purely model-generated verification evidence and require a named human to attest test adequacy, a signature-bearing tier appears in medical, avionics, automotive (ISO 26262) and now AI-product QA.
Formal release-gate ownership: SRE/change-management policy or SOX-style ITGC audit requiring a named non-developer to sign release go/no-go and own the rollback decision, distinct from the engineer who wrote the code. Watchable signal: change-advisory-board records naming QA as approver, or FDA/notified-body audit findings citing lack of independent release sign-off.
Physical device-lab and hardware-in-the-loop testing — real handsets, wearables, medical devices, vehicle HIL rigs, accessibility testing with screen readers and assistive hardware — cannot be simulated for certification. If certification bodies (e.g., automotive HIL requirements, FDA device usability testing under IEC 62366) continue to require on-hardware evidence, the physical share of the surviving role rises.
The limit. Even with all of these, the ceiling is low and narrow: the liability and embodiment routes exist only in safety-critical verticals (medical, avionics, automotive, defense) that employ a small fraction of the 186,740. For web/SaaS/enterprise-app QA — the bulk of the occupation — there is no plausible license, no signature requirement, and no route to a trust premium: buyers of software never see or ask about the tester. The lever set describes survival of a senior tier, not of the occupation's volume.
| New York-Newark-Jersey City, NY-NJ | 11,460 | $130,540 +25% |
| Dallas-Fort Worth-Arlington, TX | 8,840 | $103,210 -1% |
| Washington-Arlington-Alexandria, DC-VA-MD-WV | 8,460 | $126,350 +21% |
| Los Angeles-Long Beach-Anaheim, CA | 7,200 | $122,210 +17% |
| Seattle-Tacoma-Bellevue, WA | 6,650 | $129,690 +24% |
| San Jose-Sunnyvale-Santa Clara, CA | 6,480 | $166,530 +60% |
| Boston-Cambridge-Newton, MA-NH | 6,050 | $122,210 +17% |
| San Francisco-Oakland-Fremont, CA | 5,880 | $134,630 +29% |
| San Jose-Sunnyvale-Santa Clara, CA | 6,480 | $166,530 +60% |
| Lexington Park, MD | 400 | $147,520 +41% |
| Boulder, CO | 500 | $136,410 +31% |
EA's CEO stated that AI is used in about 85% of the company's game quality assurance work, while claiming QA headcount has increased.
Has AI actually changed your work? One tap, anonymous, and the running tally is public. Nothing else is asked of you.
Rather than check back: get the digest and we'll tell you what changed — or watch a single occupation from its own page.