Every time I participate in AtCoder Beginner Contest, when I turn to check the standings, I'd always be shocked!! This time in the first page, more than half are AI cheaters. It seems that AI cheaters are getting more rampant! Of course I know the AtCoder official is always banning those users after the contest, but there are no signs of the phenomenon of AI cheating decreasing, how???
By the way, I'm curious about why is their AI so intelligent? When I get stuck in a problem (not in a contest, just daily training), sometimes I turn to AI for help, but my AI is always so stupid.
some people keep doing things like that just to "have fun". officials should ban their ip.
Oh that's a great idea
I think problem setters should be more creative and make problems that can only be solved by a human..
As I now, ChatGPT 5.2 Thinking(xhigh) can solve G in <20min, but this is not free
This is even not 5.4 :(
comp prog getting pay to win was not expected :cry:
it will still fall for white text no?
I think it's because of two things. The first one is that some people just like to cheat. You can talk to them and try to explain to them why it's not cool to cheat (and how it's not even beneficial to them, usually) but it's in one ear and out the other for them. The second is that, since AI is so good now, the cheaters get concentrated at the top. So even if there are only $$$100$$$ cheaters, you will maybe see like $$$50$$$ of them in the top $$$100$$$.
Luckily, at least for now, cheaters' rating graphs usually look unrealistic, for example like this:
Ginge
1 year to International master!
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣀⡤⠤⠆⠐⠂⠀⠀⠀⠰⠤⠤⣄⣀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀ ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣠⡴⠖⠋⢁⣀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠠⠤⣀⡀⠉⠑⠲⢄⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀ ⠀⠀⠀⠀⠀⠀⠀⠀⠀⣠⠖⠋⢁⣠⠖⠛⠉⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠉⠳⢶⣄⣀⠉⠳⢤⡀⠀⠀⠀⠀⠀⠀⠀⠀ ⠀⠀⠀⠀⠀⠀⢀⣴⠋⠁⣠⠖⠋⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠙⠓⢄⡀⠙⢦⡀⠀⠀⠀⠀⠀⠀ ⠀⠀⠀⠀⠀⣰⠋⠀⣠⠞⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠙⣄⠀⠙⢦⡀⠀⠀⠀⠀ ⠀⠀⠀⢀⡞⠁⢀⣴⠃⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⢷⡀⠀⠱⡄⠀⠀⠀ ⠀⠀⢠⠏⠀⢀⡾⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⢳⣄⠀⠹⣄⠀⠀ ⠀⢠⡟⠘⠉⠉⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⡀⠀⠀⠀⠀⠀⠀⠈⢹⠀⠙⡄⠀ ⠀⡼⠁⠀⠀⠀⠀⠀⠀⠀⠀⠿⣿⣷⣶⣤⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣤⣶⣾⣿⣿⣂⣀⠀⠀⠀⠀⠀⠈⠃⠀⠱⡀ ⢠⠇⠀⠀⠀⠀⢀⣠⡶⠟⠛⠛⠛⠿⣿⣿⣿⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣴⣿⣿⣿⠟⠋⠉⠉⠉⠙⢷⣄⠀⠀⠀⠀⠀⠀⡇ ⣼⠀⠀⠀⠀⢀⡿⠋⠀⠀⣀⣀⠀⠀⠘⢿⡟⣿⣷⡀⠀⠀⠀⠀⠀⠀⢠⣿⣿⠏⣿⠁⠀⢀⣠⣤⣀⠀⠀⠹⡇⠀⠀⠀⠀⠀⢽ ⡟⠀⠀⠀⠀⢸⡇⠀⠀⢸⣿⣿⣷⠀⠀⠘⣷⠈⠻⣷⠀⠀⠀⠀⠀⠀⣿⠟⠁⢸⡇⠀⠀⢸⣿⣿⣿⠀⠀⠀⣿⠀⠀⠀⠀⠀⠈ ⡇⠀⠀⠀⠀⠸⣧⠀⠀⠈⠛⠛⠋⠀⠀⣰⣿⡢⠀⠀⠀⠀⠀⠀⠀⠀⣀⢴⣻⣿⣷⡀⠀⠀⠉⠋⠉⠀⠀⣰⠏⠀⠀⠀⠀⠀⢠ ⢹⠀⠀⠀⠀⠀⠘⢷⣄⡀⠀⠀⠀⣀⡴⠋⠙⠿⠳⢤⣀⡀⠀⠀⠀⣼⣷⣿⠟⠁⠈⢿⣦⣀⡀⠀⣀⣠⠾⠋⠀⠀⠀⡀⠀⠀⠀ ⠈⣦⣤⡾⠶⠶⠖⠒⠛⠛⠿⣶⣿⢿⣤⣤⣤⣤⣤⣦⣿⣷⣤⣤⣶⠏⠉⠀⠀⠀⠀⣼⡿⢷⣿⣿⣿⣶⣶⣶⣶⣶⣾⠛⢷⣄⠄ ⠀⣼⠏⠀⠀⠀⠀⠀⠀⠀⢰⢿⡇⠀⠀⠀⠀⠀⠀⠉⠉⠉⠉⠉⠉⠀⠀⠀⠀⠀⠈⠛⠃⠀⠀⣰⢶⡉⠉⠉⠉⠉⠉⠀⠀⢻⡄ ⠀⡏⠀⠀⠀⠀⠀⠀⠀⠀⠋⣸⣷⣶⣤⣀⣀⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣀⣀⣀⣤⣶⣾⣿⡀⠀⠀⠀⠀⠀⠀⠀⠀⠈⣷ ⠀⣷⠀⠀⠀⠀⠀⠀⠀⠀⢠⣿⡏⠉⠉⠛⠛⠛⠛⠻⠿⠿⠶⠶⠶⠾⠿⠿⠟⠛⠛⠋⠉⠁⢈⣿⣧⠀⠀⠀⠀⠀⠀⠀⠀⢠⡿ ⠀⠸⣧⡀⠀⠀⠀⠀⠀⣠⡿⠉⠻⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣴⠟⠁⠘⢷⣄⡀⠀⠀⠀⠀⣠⡿⠁ ⠀⠀⠈⠙⠲⠤⣤⠶⠟⠋⠀⠀⠀⠈⠙⠷⣤⣀⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣀⣤⠾⠋⠁⠀⠀⠀⠀⠙⢿⣷⡶⠖⠛⠁⠀⠀ ⠀⠀⠀⠀⠀⠀⠈⠓⢤⡀⠀⠀⠀⠀⠀⠀⠀⠉⠙⠛⠲⠦⠤⠤⠤⠤⠖⠚⠋⠉⠀⠀⠀⠀⠀⠀⠀⢀⣴⠞⠁⠀⠀⠀⠀⠀⠀ ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠙⠢⣄⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣠⠶⠋⠁⠀⠀⠀⠀⠀⠀⠀⠀ ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠉⠒⠤⢀⣀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣀⣠⠶⠛⠉⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀ ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠉⠓⠒⠲⠤⠤⠤⠴⠦⠔⠒⠛⠉⠉⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
What motivates AI cheating to stop?
If you don't have an answer to that — you have an answer to your question.
I have the same experience. Very hit and miss.
AI solutions are much less reliable than people think, especially on higher-division problems. Most models generate plausible looking code, but it usually fails on tricky corner cases or misses the key insight. If someone doubts that, try asking an AI to solve recent Div1 E/F/G problems consistently.
If people are curious about how strong competitors actually approach these problems, watching post-contest upsolve streams can be helpful — for example streams from TLE Eliminators or Shayan, where solutions and thinking process are explained.
As for my rating graph: I practiced a lot offline and solved many problems virtually before participating in rated contests. That’s why the initial growth looks steep. I’ll keep participating in upcoming contests, so the rating will stabilize naturally over time.
There are also multiple players in the current top-100 who reached IGM in very few contests, so fast progress by itself isn’t unusual.
funny that this guy is cheater.
I'm so sad about the cheating situation. This platform means a lot to me because it's where I began learning CP
aren't u a cheater too?? ur last contest submissions look sus 335394219