Mechanical engineer

United States. The statistics bureau projects that employment of mechanical engineers will grow 11 percent from 2025 to 2035, much faster than the average for all occupations, with about 17,800 openings a year. It says employment is expected to increase as automation and innovation create demand for these workers, and that as manufacturing processes incorporate more complex automation machinery, mechanical engineers are expected to be needed to help integrate this equipment into existing systems. It counts jobs for one country, not tasks.

A named person with standing publicly predicted something, on a date, in an attributable statement. It is recorded so that who said what, and when, stays checkable — and it never moves a task's assessment, because a prediction is not an observation. Its value arrives later: the record sits on the same page as the evidence about that occupation, so anyone reading the forecast reads the record of what happened next beside it. That is the reckoning; this site publishes no verdict on whether a forecast came true.

A benchmark of engineering problems that models solve by driving COMSOL Multiphysics, a commercial finite-element tool, through its API. The best strategy generated executable API calls 88% of the time; but the paper says the model passed the strict cut — a valid target within 10% of the correct answer — for a single problem in two of its experiments, and on the gold problems Claude 3.5 Sonnet produced a valid target for 1 of 15. It is a small preprint benchmark with 2024–2025 models, and the authors' employer develops one of the models tested.

A demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.

An academic benchmark of 200 natural-language prompts paired with expert-annotated CAD scripting code. It reports that GPT-4 produced code that compiled 96.5% of the time, compared with 85% for Gemini and 73.5% for CodeLlama, and the authors note that the 3D object a generated script renders may not meet the specified requirements, which is why they propose a verification step. It measures whether scripts run and resemble the target, not whether parts are manufacturable or correct.

A demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.

A benchmark built on the Formula SAE student competition's rule document, with CAD images and engineering drawings from the MIT Motorsports team. It reports that the models tested, while promising, struggle to reliably retrieve relevant rules, face challenges in recognising technical components in CAD images and encounter difficulty in analysing engineering drawings; even given the full rule document, GPT-4o could not reliably extract specific rules. It tests 2024 models on one competition's rules, and several authors work for a design-software company.

A demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.

European Union. The regulation lists, in Annex I Part A, safety components with fully or partially self-evolving behaviour using machine learning approaches ensuring safety functions, and machinery with such embedded systems; for these the manufacturer shall apply a third-party conformity assessment procedure. Its essential requirements say risk assessment shall include hazards foreseeable as an intended evolution of the machine's self-evolving behaviour. It applies from 14 January 2027. It binds manufacturers and governs how machines are assessed, not whether engineers' tasks are automated.

Regulation, subsidy or public procurement is requiring or funding adoption — the mirror of a constraint. It shows adoption is being required, not that it has happened, so one mandate is never enough on its own; two independent ones are.

NASA Goddard Space Flight Center, one engineer's programme using commercially available generative-design software. A CAD specialist starts from the mission's requirements and draws the surfaces where a part connects; the software then produces complex structure designs in as little as an hour or two, and the engineer says the evolved parts save up to two-thirds of the weight of traditional components. He also says the algorithms need a human eye because they can make structures too thin, and the parts are analysed with NASA-standard validation software and processes. The figures come from the engineer who pioneered the method, and it is one organisation's use.

An employer has put it into production. Can move the baseline — weighted by scale and how similar the setting is.

Texas, United States. The statute says a person may not engage in the practice of engineering unless the person holds a license, and that a plan, specification, plat or report issued by a license holder for a project to be constructed or used in the state must include the license holder's seal. It exempts employees of a private company working only on the research, development, design, fabrication, production or service of products the company manufactures — so many product-design mechanical engineers need no licence. It says nothing about software.

Failure, rollback, regulation or cost is suppressing adoption. Can lower an assessment or widen its uncertainty.

If you are starting out, expect generative tools and AI assistants to draft parts, scripts and simulation set-ups, and expect the checking to be yours. The durable skills are specifying what a part must do, judging whether a simulation's number is credible, and testing real hardware — and, in regulated work, the licence to seal designs.

Expect generated geometry and simulation drafts to arrive faster than you can check them. Your value moves towards constraints, validation and safety: EU machinery law will require assessments of machine-learning safety functions from 2027, and factories adding automation need engineers to integrate it.

Stay and strengthen

Stay in design, and own the constraints and the validation

Generative tools need someone to define the constraints and catch parts that are too thin; simulation agents rarely reach a correct number unaided.

Reshape the role

Move towards machine safety and conformity assessment

From 2027, EU law requires third-party assessment of machine-learning safety components and risk assessment that covers how they evolve — work that needs engineers who understand both machines and software.

Adjacent move

Move into manufacturing automation and integration

US projections expect engineers to be needed to integrate more complex automation machinery into existing production systems.

Not on present evidence. Generative design and AI assistants draft parts and scripts, but benchmarks show models struggling to read engineering drawings and rules and rarely reaching correct simulation answers unaided; the EU will require third-party assessment of machine-learning safety components from 2027; and US projections expect the occupation to grow 11 percent from 2025 to 2035.

We do not answer that with a number of years. A signal to watch instead: whether AI tools start delivering parts and simulations that pass validation without an engineer changing them, and whether safety law stops requiring people to assess and sign. Today generated parts are still validated by people, and the EU is adding assessment duties, not removing them.

Generative design software proposes a part's shape once an engineer has set the requirements, connection points and keep-out zones. NASA reports it producing complex structures in as little as an hour or two, but says the algorithms need a human eye, and the parts are still validated before use.

US projections expect employment to grow 11 percent from 2025 to 2035, much faster than average, partly because manufacturers adding automation need engineers to integrate it. The work shifts towards specifying, validating and safety.