ChatGPT solved many of the AoC 2022 problems with no assistance, and those are written in an even more obfuscated style than GCJ. I imagine the LLMs of this summer and next summer would blow through most of the early rounds of GCJ with ease.
AoC tasks are much easier than usual Competitive Programming tasks. AoC is pretty much undergraduate class level (whereas some Competitive Programming problems ended up being turned into actual papers) and the main difficulty in tasks is in implementing the solution rather than coming up with a tricky idea. Also, in AoC time complexity doesn't matter at all - all you need is any solution. Classic Competitive Programming problems have both a memory and a time limit. Most of them have a simple brute force solution but the trick is finding an efficient algorithm.
However, I can see Google execs thinking the same and using this explanation as the reason to close the competition.
