Should I trust AI for engineering design and simulation?

You either ignore it or trust it blindly, and neither is the judgment an engineer is supposed to bring. The tool is not a thing to believe in or dismiss. It is fast and confident and has no idea whether its answer is true, and your job is to be the one who does. Use it where mistakes are cheap and checkable: exploring options, rough passes, speeding up the tedious parts. Hold it at arm's length where a mistake is expensive or hidden, the core design, the safety-critical calculation, anything real people will depend on. A simulation's confidence tells you nothing about whether its assumptions were right. Only your validation does. The tool proposes; you are the one who verifies and signs off. So turn it from a mood into a method. Before you accept any output, ask how bad it is if this is wrong and whether you would catch it, then check accordingly. Let the stakes decide your trust, not your comfort with the tool. That is what engineering judgment looks like now.