Challenge: Can code AIs fix real-world problems?

On this site, some programmers get nervous about coding AIs taking away their jobs.

Meanwhile, some of the most basic bugs that plague millions of programmers are still around after decades.

If code AIs were really good, shouldn't those issues disappear?

So here's a challenge:

* Full-text search in gnome-terminal does not work; it misses occurrences (https://gitlab.gnome.org/GNOME/vte/-/issues/220)

If your coding AI, or your use of somebody else's coding UI, can convincingly fix this, please post!

I also invite others to post similar challenges of "this is so basic, how hard can it be?" style problems in real-world open-source software that lots of people use.