The same way I verify that my own regex is correct: by running a few test cases and then crossing my fingers.

Just like programming tricks in any language supported by copilot:

a. Code, once provided, can be broken down and understood at a far easier level than is required for composition;

b. Worst case, try several test cases to both increase comprehension and reduce the chance of 'gottcha's.

Shouldn't be too hard to stick with option 'a' as clear best practice, looking up any operators or syntax that aren't immediately obvious, the advantage being that the AI can use obscure tricks that you aren't initially aware of but you still have the opportunity to review and understand the regex, becoming better over time. It's theoretically auto-generated, but practically computer-assisted.

That is the opposite of what most people will say. To most, reading someone else’s code is much harder than constructing their own. Try this one:

    @P=split//,".URRUU\c8R";@d=split//,"\nrekcah xinU / lreP rehtona tsuJ";sub p{

@p{"r$p","u$p"}=(P,P);pipe"r$p","u$p";++$p;($q*=2)+=$f=!fork;map{$P=$P[$f^ord ($p{$_})&6];$p{$_}=/ ^$P/ix?$P:close$_}keys%p}p;p;p;p;p;map{$p{$_}=~/^[P.]/&& close$_}%p;wait until$?;map{/^r/&&<$_>}%p;$_=$d[$q];sleep rand(2)if/\S/;print

Smartass answers only. Ask autopilot to write a proof with --nojargon so juniors will get it! Write a single unit test and call it good! Step through it in a debugger on that one unit test to be sure? Sure of what? I dunno but it sounds diligent...

When I watched Idiocracy, a small optimist in me said "but surely the techies..." That optimist has died. We're fucked

Probably the way most people do - they run it over whatever examples they can think of at the moment as a check, and then forget about it till it breaks.