AI
AI for AEC
Applications that can be checked. Limits and human review stay in view. Articles appear here as they are published.
This is not a substitute for engineering judgment. A language model is not an authority on design, analysis, or safety.
Draft OpenSees input with AI — then prove it with a solver and a human
A language model can sketch OpenSeesPy. Only a reviewed model, a real analysis run, and an independent check produce engineering evidence.
An LLM is not a solver
Separate deterministic calculation, AI suggestion, human review, and validation evidence. A language model can draft text. It does not occupy responsible charge.
Retrieve from manuals. Do not invent code clauses.
Retrieval-augmented generation can point a reader at a real manual. It still does not write law, and it must not fabricate clause numbers.