Antenna design is not black magic. There are a few well-understood basic patterns and you parametrize those. Some of those look quite exciting and tuning the matching network is annoying if you've never done it before but typically you use a template for an antenna design and set the parameters to what gets you best simulation results, then prototype and measure and set the matching network to match your measurements. It's not magic, it's normal everyday engineering.

Coming up with new fundamental antenna designs feels a bit black magicky to us lowly electronics people, but that's just because it's more applied physics than engineering. Here's an example of how new antenna pattern design is done: http://sci-hub.tw/https://www.sciencedirect.com/science/arti...

No it's not, it just seems that way to the vast majority of us who haven't developed the proper intuition.

Of course, there is the GA evolved antenna.

https://en.wikipedia.org/wiki/Evolved_antenna

Slightly related, an FPGA circuit designed by a genetic algorithm which ended working due to analogue effects and hardware-specific magnetic flux interference.

https://www.damninteresting.com/on-the-origin-of-circuits/

It makes you wonder if it would be useful to create a programmable circuit where such analogue effects are the intended working principle.

And the 'thing' device http://www.cryptomuseum.com/covert/bugs/thing/index.htm

For this or other successful genetic algorithms, it would seem clear you still can’t know if it’s an optimal design simply because in most cases the number of designs tested would be a small fraction of the possible designs.

However, things like useful sound recognition being done with only a small number of logic gates (commenter below provided a nice article, thank you) make it hard to imagine doing much better.

I wonder if the process can be shown theoretically to offer any help in guaranteeing minimum bounds w.r.t. the optimal case, even if can’t be fully proven to be optimal.