Richard Johnson

Stop letting a chat AI produce your engineering numbers: it predicts tokens, not computes. Use AI to set up and explain, a computational engine to solve, and verify.

When an engineer should deliberately not use AI: stamped work, life-safety calculations, confidential data, unverifiable output, novel problems, and code work.

Compare candidate parts’ datasheets fast without picking wrong: let AI tabulate and normalize, then confirm every deciding value on the primary datasheet.

AI CAD software is priced four ways: AI bundled in the subscription, at a higher tier, by usage or credits, or by enterprise quote. Verified 2026 prices.

Which AI tool should an engineer use? A decision framework: pick by task, discipline, data sensitivity, and budget, then verify every output.

AI speeds up datasheet search and component selection by building a shortlist, but part numbers, stock, and lifecycle need confirming on a distributor.

AI tool terms of service decide who owns outputs, who is liable, and whether you are defended. The six clauses engineers should check before adopting a tool.

AI can help read a datasheet by summarizing and explaining, but is unreliable at extracting exact values from tables, ratings, and graphs, so verify every value.

Using AI with a client drawing set is safe only on a no-training business tier, backed by a data processing agreement, with client consent.

Can AI do CAD? It assists in four real ways but does not autonomously produce production-ready editable parametric models; the engineer stays in charge.

AI can explain wiring-diagram symbols and interpret small circuits, but it is unreliable at extracting exact connections from a dense schematic.