Human safety adds massive overhead to any mechanical engineering project. I worked in aerospace. In probably 1-2 days, we could conceivably change a part drawin

Human safety adds massive overhead to any mechanical engineering project. I worked in aerospace. In probably 1-2 days, we could conceivably change a part drawing, send it to the shop, then have it start being manufactured en masse. Why don't we do that? Because the part drawing must be reviewed by the analysis group first. Altering part geometry can change load paths. We might have enlarged something to strengthen it, but now it requires a larger hole elsewhere. It can take days to months to anaylze the part change. We do manual calculations, physical simulations, and physical testing. Each of those is still not at the point where we could pass it to a strong AI, they're all part art and part science that require an engineer's manual, informed decisions. Correcting experimental data, deriving new calculations, even just making diagrams and describing all of this along the way takes time. We must document everything. We might have to design and build a test fixture from scratch (airplane wing test fixtures aren't exactly off the shelf). Experiments get run wrong, they set the wrong loads, a tech installed a sensor 1 inch off, meaning we have to throw out a month's worth of data. Humans make mistakes. We do our best to catch them. Only after all this can we get regulatory approval (FAA in our case) and start pushing it to production. Honestly the attitude that software eng is "light years ahead" is infuriating, mechanical engineering is rife with time consuming checks and triple checks so that people do not die.