Question of the Month: Would you trust an AI to select components for your next system design?

I would use AI to find and point out the highest tech qualities available at the time and then get its opinions for what the best are based on price, actual availability and intention for purpose.. I would definitely use the information learnable from AI to make my decisions but I don't think I would outwrite a while AI to choose all of the components itself and make the decisions itself I just don't know if I'm there yet

I selected "Other" - it would completely depend on the input I was able to give the AI. If I can give it all the inputs I would normally use when filtering, then ask it to optimise a list by some factor(s) that could potentially give me a good shortlist. If it can then go through the datasheets of every part in the shortlist for all the parameters/operating conditions that don't make their way to filters it would save me a lot of time ruling out parts based on datasheet values. If the AI can then go and run spice models to determine the optimal part and give me a report on all the results and why it sees a specific part as optimal based on my inputs, anything I might have missed, then I'd be more inclined to take its input into account or select parts based on its feedback.

Basically it all comes down to how much data the AI has access to, and how much effort it will put in to select a part. Otherwise you may as well just ask for peoples suggestions on forums/discord if you have no idea where to start. I can see it saving a lot of time if implemented well, and causing a lot of frustrations if implemented poorly.

My favorite part of the article I cited is where the lawyer asks ChatGPT if it's a real case, and ChatGPT says yes and none of the other citations are fake either!

In more ways than one it it would appear...

" ...The lawyer who used the tool told the court he was "unaware that its content could be false"... "

https://www.bbc.co.uk/news/world-us-canada-65735769

AI doesn't appear to be excelling itself on the automated scanning front either by the looks of it... 

" ...AI scanner used in hundreds of US schools misses knives... "

https://www.bbc.co.uk/news/technology-65342798

I know so little about AI I wouldn’t know what to think about it.

A word of caution:

what happened when a lawyer used ChatGPT

True, there are solution spaces where Augmented Reality can be tied to a limited AI discriminator to provide useful answers to very specific questions.

It will be a long while before I could trust an AI solution in human safety domains.

There has to be a middle ground. Rarely does a new technology start with the most risky implementations.

I bet it can select parts for a gizmo 

Would you trust AI to drive you in a car, or drone-deliver your purchases? These questions really boil down to not if but when will you trust AI do do important work?  And then when will you start not trusting it?  As in, when do you think Skynet will not have your best interests at heart? 

Personally I would allow the AI to do a baseline list of upgradable components then I would wouldndouble check everything to make sure it's what would be best for the system. More of a checks and balances system than giving it free reign