AI systems can be impressive with text and images, but architectural drawings create specific challenges. Drawings are not just pictures. They are layered technical documents that depend on scale, callouts, sheet references, schedules, notes, sections, details, and jurisdiction-specific requirements.
One common weakness is spatial reasoning across sheets. AI may recognize a plan, but still misunderstand how a section, elevation, detail, or callout relates to it. It may also miss conflicts between a written scope, a detail note, and a product document.
Another issue is confidence. AI may produce a polished explanation even when it has not truly verified the drawing set. In professional workflows, a confident answer is not enough. The answer has to be traceable to the drawings, code data, documents, and applicable requirements.
AI can still be useful when it is constrained and verified. It can help find information, draft summaries, and organize review questions. But final decisions about code compliance, safety, and document coordination must remain with qualified humans using the full project context.