Hi! This is a big interest of mine as a CTO who spans both HW and SW. LLMs are really good at things where creation is hard but validation is easy. That's a great fit for software where unit testing is easy, but a big challenge for hardware where "validating" can take weeks for a PCB to come back from the fab.
I've been playing with GPT4 for both software and hardware and I've been impressed with how much electrical background knowledge GPT4 has, but because of the validation problem it doesn't "just work" out of the box like it does with software.
There are a lot of interesting places to add it in the hardware value chain though :) Would love to chat more about it if you're interested!
1) What particular issues do LLMs have with at the validation stage? RF and EMI are ones I can see as issues, but I would think the digital logic portion could easily be solved.
2) Seeing as you run a robotics company, do see any notable applications so far to computer vision or control systems within the near future?
2 comments
[ 3.1 ms ] story [ 18.1 ms ] threadI've been playing with GPT4 for both software and hardware and I've been impressed with how much electrical background knowledge GPT4 has, but because of the validation problem it doesn't "just work" out of the box like it does with software.
There are a lot of interesting places to add it in the hardware value chain though :) Would love to chat more about it if you're interested!
1) What particular issues do LLMs have with at the validation stage? RF and EMI are ones I can see as issues, but I would think the digital logic portion could easily be solved.
2) Seeing as you run a robotics company, do see any notable applications so far to computer vision or control systems within the near future?