Good luck with AI designing anything related to a pcb. Think about all the aspects and process involved.
Imo a autorouter is sort of AI but it requires guidance (control) or it has no idea what you want as a outcome. A router needs parts placed, routing rules, guidance to be effective.
Okay AI, create a kicad footprint for a Omron LY-01 relay please, here is the PDF spec sheet. Just something I had to do the other day.
Imo a autorouter is sort of AI but it requires guidance (control) or it has no idea what you want as a outcome. A router needs parts placed, routing rules, guidance to be effective.
Okay AI, create a kicad footprint for a Omron LY-01 relay please, here is the PDF spec sheet. Just something I had to do the other day.
If it's as good as the code I routinely see generated by AI, I'd approach with extreme caution. Which doesn't mean it'll be that way in 5 years time, I fully expect to see AI layout in commercial systems and fairly soon.
It might be quite interesting to see the effect it has on simulation/emulation and the device models they use.
It might be quite interesting to see the effect it has on simulation/emulation and the device models they use.
I would not hold my breathe on AI designing a pcb even 5 years from now. It comes down to the abstraction layers. How do you define its control? Okay I want a audio power amplifier, it’s all in the details which can become enormous and so many variables to be defined. Talk about an engineering spec. You might take more time writing a spec in a controlled way than actually doing the work to get it done.