Aerospace Engineer

Aerospace Engineer scores 12/100 on this site's structural exposure measure, which puts it in the Higher Risk band. That number is not a quirk. Concept studies, aerodynamic and structural analysis, requirements documents and test reports are computer-mediated, technical-language work, which is exactly what the OpenAI and AIOE pillars rate as exposed. The research verdict is "AI-resistant despite the paperwork" at Medium confidence, and the reasoning is narrower than for some other engineering jobs here.

The protection is not a personal license. Civil and environmental engineers are shielded by the Professional Engineer stamp, but BLS says licensure is not required for entry-level aerospace positions and a PE is optional, earned by experienced engineers who want to sign off projects. Aerospace is protected differently, through certification. Under 14 CFR 21.

33, the applicant must carry out the inspections and tests that show an aircraft, engine or propeller meets the airworthiness requirements, and must allow the FAA to run its own flight and ground tests. Under 14 CFR 183. 11, the FAA's Aircraft Certification Service may select designated engineering representatives from qualified applicants, and 14 CFR 183. 29 sets out their categories, such as structural, power plant, systems and equipment, and flight analyst.

They approve engineering data within limits prescribed by, and under the general supervision of, the Administrator.

AI is already in the workflow, and it is worth being specific about where. An Aerospace America feature of 1 October 2026 reports that GE Aerospace used generative AI to produce a preliminary hypersonic ramjet design in seconds, a task that normally takes months, and that GE uses it only in early-stage design.

Airbus announced a partnership with Mistral AI on 28 May 2026 whose named uses include AI-driven simulation to optimize parts and "ad-hoc support to engineers" through development, test and certification. NVIDIA reported in November 2025 that Northrop Grumman built an AI surrogate for nozzle simulation on Luminary Cloud's platform, intended to let engineers explore thousands of designs quickly. That last item is a stated expectation, not a measured result.

Regulators are keeping a human accountable. According to the same Aerospace America article, EASA requires human oversight and does not currently allow one AI tool to verify another's preliminary design, while the FAA treats AI as a tool within its existing framework and lets applicants propose their own means of compliance, subject to FAA acceptance. Both regulators favor starting with non-safety-critical uses.

The article also notes that regulators expect engineers to show how each part was designed, and a Rolls-Royce executive quoted there says current AI tools cannot meet that standard. Engineers still run the physical component and engine tests and build the compliance record.

What this does not show matters just as much. A certification wall protects the sign-off and the test campaign, not the number of analysts needed to prepare them. If simulation becomes ten times faster, a program may need fewer people doing routine analysis, and entry-level analyst work is the most plausible place for that to appear. We searched for a named case of aerospace engineers being cut because of AI and found none.

This is an absence of evidence, not proof of safety.

The demand data points the other way for now. BLS projects employment to grow from 68,700 in 2025 to 74,400 in 2035, a gain of 5,700 jobs and 8%, against 3% for all occupations, with about 3,800 openings a year. The median engineer earned $134,960 in May 2025. In practice the engineers best placed are those who can check an AI-generated result against physics, defend it to a regulator and run the test that proves it.