What do you think about these machine learning-based functionalities?

[magnushp]

I'm doing a project on machine learning in CAD. Therefore, I'm interested in hearing your opinion on the functionalities that ML research might unlock in the future.

I have attached 7 ideas that are built on the questions:
What can be done, when the system is able to automatically detect fasteners in an assembly?
What can be done, when the system is able to infer the assembly structure?

Feel free to comment on whichever idea you'd like. E.g. would you find it useful? Do you think it would even be possible?

Also, if you have some general opinion on the potential of suggestive modeling in CAD, I would love to hear it

Ideas:
(Google Drive - I had issues uploading it here)

[matt]

I think in order for CAD to go anywhere meaningful with AI, there are going to have to be some base design scenarios or part/assembly functionalities established and well understood. Things like types of connections between parts, relationships between parts, manufacturing processes, understanding loads, some sort of styling input, target weight/ strength/ life ...

AI could reduce some of the programming needed for engineer to order, configurable or modular design. To get past that, it's going to have to develop a lot past where it is now.