CAD, AI, and your Innovative Brain
Michael’s experience illustrates a potentially profound change in mechanical engineering: the movement from computer-aided design (CAD) toward AI-directed engineering design.
What makes this example remarkable is not simply that ChatGPT generated a 3D-printable replacement part. It is that the engineer was able to communicate design intent through photographs, measurements, and ordinary language rather than manually constructing geometry inside CAD software.
However, there is an important distinction. The engineer did not disappear from the design process. His role changed. He supplied the dimensional constraints, recognized mistakes, directed revisions, and validated the physical result through test prints. The AI performed much of the geometric construction.
For Tharaka Invention Academy (TIA), I see three especially important implications.
First, the technical barrier to invention may be falling dramatically. An apprentice who understands a mechanical problem but lacks years of CAD experience could potentially develop and test physical solutions much sooner. This is especially significant for communities where access to specialized engineering software, formal training, and professional designers is limited.
Second, the value of engineering judgment may actually increase. AI can generate a convincing-looking model without understanding whether the component will survive repeated loading, tolerate sunlight, fit its mating parts, or fail safely. The cargo net clip might look nearly identical to the original while having substantially different mechanical properties because of the resin, printing orientation, or internal structure.
A successful replacement print is therefore evidence of geometric and functional feasibility under the conditions tested, not proof of engineering equivalence.
Third, invention education may need to change its priorities. Instead of beginning with months of software instruction, TIA could introduce apprentices to a workflow in which they identify a problem, document the original component, establish dimensional and functional requirements, use AI to generate a model, fabricate a prototype, and evaluate its performance.
That approach fits TIA’s emphasis on learning through invention rather than mastering tools as an end in themselves.
There is, however, a deeper question raised by Michael’s reaction.
If AI increasingly performs the technical tasks traditionally associated with engineering, what should an aspiring inventor actually learn?
I would argue that apprentices still need a working understanding of geometry, materials, mechanics, manufacturing processes, measurement, and experimental validation. What may become less essential is manually executing every intermediate drafting operation.
This also suggests an opportunity for TIA’s Problem Pals™. An AI Mechanical Design Companion could guide an apprentice from photographs and basic measurements through geometry generation, printable file preparation, prototype evaluation, and iterative improvement. It would need to distinguish clearly between AI-generated suggestions and physically validated results.
The most interesting educational possibility is not that everyone becomes an engineer simply by describing an object to AI. It is that many more people could begin practicing invention while progressively developing the engineering knowledge needed to make their designs reliable.
I’m more interested in exploring whether ChatGPT can actually perform this kind of CAD-to-3D-print workflow today?
(2)
