🎭 IJCAI 2025 Review Results Are Out! Let the Gossip Begin...
-
It's that time of year again. The IJCAI reviews have dropped. The salt has flowed. The memes are writing themselves.
IJCAI 2025, one of the most anticipated (and controversial?) conferences in AI, has just released its review results. And oh boy, did it live up to its reputation. From reviewers asking “What is Softmax?” to borderline rejections over missing “ethical considerations” no one knew were mandatory... the drama is real.
🧂 Initial highlights from the community, so far ...
“Got a reject. Only comment from the reviewer: ‘What is softmax?’”
— Zarachoadr, living meme legend.“Reviewer asked why I didn’t report precision@1 and recall@1... on a generative model.”
— ahfyj, somewhere crying in PyTorch“Was told my method may not be reproducible. I attached the code and even said ‘just pip install torch and dgl’.”
— ahfyj, again, deserves a hug“Got rejected for not including ‘ethical considerations’. I didn’t see that as a required field?”
— wwww, bewildered“3 Accepts, 1 killer WR. Now I don’t even know if rebuttal is worth it.”
— 清茶品也醉, currently coping“I swear IJCAI is pure lottery. Top-tier randomness. Still hurts when the slot machine shows REJECT.”
— 凔河, realist philosopher
Got Your Own Story?- A reviewer clearly misunderstood your domain?
- Got full scores but got rejected anyway?
- Miraculously accepted with 3 borderline reviews?
- Still waiting for that “maybe we’ll add a 3rd reviewer” email?
️ Drop your war stories below. Screenshots, spicy quotes, spicy takes.Let's make this thread the unofficial after-party of IJCAI 2025 reviewing season.
Rebuttal Tips -- or, is it just therapy now?Still planning to rebut? Here’s what we think we know:
- 1-page rebuttal PDF only.
- You cannot change the paper content.
- Channel your inner Zen master. Or roast lightly. Your call.
May your rebuttals be read and your memes be shared.
Let’s talk it out, flame it out, and maybe — just maybe — laugh it out.
🧵 Register (verified or anonymous) to Join the Discussion
-
I collected some community reported scores (from Zhihu, Reddit and Linkedin) to share with all of you; and feel free to add to it.
# Score Combo Verdict Reviewer Comment Highlights Mood 1 SA, CA, WA, WR Who knows “Reviewers don’t read code. Or stats.” 
️2 WA, BR, WR Probably RIP “Why no precision@1 and recall@1… on a generative model?” 
3 3x WA, 1x BA Slim hope “Innovation weak + unclear experiment” 
4 BR, BA, WA Schrodinger's paper “Got 3 reviews, 1 misunderstood the entire method” 
5 2x CA, 1x WR Toss a coin “WR says paper is not reproducible despite submitted code and torch reqs” 🧂 6 1x SA, 1x CR, 3x WA, 1x WR Chaos mode “Got 6 reviews. Don’t know if that’s lucky or unlucky.” 
7 1x WR only Instant KO “What is softmax?” 
8 3x Accept, 1x WR 
“Now I don’t even know if rebuttal is worth it.” 🧘 ️9 1x WA, 2x BR, 2x WR Let it go “Reviewer: why no ethical considerations?” (not mandatory) 10 5, 5, 5 Suspense “What’s the bar here, exactly?” 
11 2 reviewers only RNG incoming “Third review popped in 5 mins before rebuttal. Instant reject risk.” 
12 Special track submission No rebuttal “Straight to judgment. No therapy PDF allowed.” 
13 3x WA, 1x BA Rebuttal Copium “Experimental section weak, but I think it's salvageable.” 
14 67 / 566 / 4567 Audio gang? No context. Just vibes. 
What Now?- Rebuttal PDF: One page, no changes to the paper. Summon your inner Zen master or lightly roast with class.
- Deadline: Blink and it’s over. AOE timing = shorter rebuttal window than your last vacation.
- Hope Level: Somewhere between “NeurIPS fallback” and “Switch to marine biology.”
🧂 Spicy Take of the Day:
“IJCAI is just a slot machine for AI researchers with a badge system.” — Everyone, probably.
Got your own combo of scores, strange reviewer quotes, or rebuttal epiphanies? Drop them below — therapy is more effective when crowdsourced.
May your WRs read your rebuttals. Or not exist at all.
-
Hear me out:
We roast IJCAI every year… and still can’t resist hitting that submit button. Some habits die hard.
-
Here are the historical acceptance rate of IJCAI:
Conference Acceptance Rate IJCAI'13 28.0% (413/1473) IJCAI'15 28.6% (572/1996) IJCAI'16 24.0% (551/2294) IJCAI'17 26.0% (660/2540) IJCAI'18 20.5% (710/3470) IJCAI'19 17.9% (850/4752) IJCAI'20 12.6% (592/4717) IJCAI'21 13.9% (587/4204) IJCAI'22 14.9% (679/4535) IJCAI'23 14.1% (643/4566) -
[IJCAI 2025 Starts the "Lottery"! An Urgent Accept Withdrawal!]
The IJCAI 2025 Program Chairs have started the "lottery" process!
Over the past couple of days, many participants noticed their submission status changing from Accept to Reject — good thing some didn't pop the champagne too early...Last year, a similar situation occurred: the scores were briefly made public for a few minutes before being hidden again. Now, it seems to be happening once more.
From what I understand, some papers, especially those rated borderline by Area Chairs (ACs), can indeed shift from Accept to Reject during final decisions.
Please note: the official outcome will be based on the final notification email. Stay tuned!
How to Check Your Result Using a CMT Bug
Here’s a method to check your submission status early by using a small trick:
-
Log in to the CMT (Conference Management Toolkit) website — simply search for it online if needed.
-
After logging in, open a new browser tab and enter the following URL (replace
xxxwith your paper ID):
https://cmt3.research.microsoft.com/api/odata/IJCAI2025/Submissions/xxx -
Replace
xxxwith your specific paper ID. -
Look at the
statusidfield:
1= Pending2= Accepted3= Rejected79= Likely to be Rejected80= Likely to be Accepted
-
-
I was forcibly assigned to review a paper that is almost irrelevant to my field.
After reading it, I found the quality to be decent, so I gave it a weak accept (changed after being reminded).
Thinking about it, the reviewers for my own submitted paper are probably in the same situation.
-
Community reports:
Software Enginnering research domain.
8 submission in total.
- 5 tobe rejected，the scores are 8773，776，8757 respectively;
- 3 tobe accepted? the scores are 75432，6655，346，4343，6633 respectively.
-
Sorry for entering a mismatched topic. But can anyone give my an answer or explain for the below question?
Have the ICLR 2025 organizers completely forgotten that they claimed on their official website that they would select high-quality reviewers and that these reviewers “will receive special acknowledgment during the opening ceremony and free registration”? They didn't do what they were supposed to do. Did they announce this in the opening ceremony?
-
Highest Acceptance Rate in 5 Years, but Confusing Rejections?
Hi there,
Just wanted to share some thoughts (and vent a bit) after the IJCAI 2025 notifications dropped last night.
A Record-High Acceptance RateThis year, IJCAI hit a 19.3% acceptance rate, which is the highest in the past five years. Here’s the trend based on the numbers reported:
Year Submissions Acceptances Acceptance Rate IJCAI 2021 4204 587 13.9% IJCAI 2022 4535 679 14.9% IJCAI 2023 4566 643 14.1% IJCAI 2024 5651 791 14.0% IJCAI 2025 5404 1042 19.3% So yeah, on paper, things are looking good for IJCAI this year in terms of openness. But…
🤯 Confusing Reviewer Decisions
Many people were left scratching their heads when they saw meta-reviews marked "Accept" but still got rejected in the final decision.
Take this example that’s been circulating online:
“What is your suggested final decision? Accept. Note that your proposed decision may be overruled by the AC or Program Chair…”
Well, overruled it was. The paper was ultimately rejected.
This has led to a wave of confusion. Some say it’s not uncommon: AC accepts, but PCs or chairs veto it.
Others argue that most top-tier conferences do this internally but don’t show the contradictory info so explicitly. IJCAI just happens to be more "transparent" about the mess.
Either way, the inconsistency between meta-reviews and actual decisions has sparked a lot of debate. Some are calling it disorganized. Others think this transparency at least gives us insight into how chaotic things can be behind the scenes.
Shoutout to everyone who did get accepted — huge congrats to you!
Anyone else experience weird decision conflicts this year? Would love to hear how your reviews went.
-
Highest Acceptance Rate in 5 Years, but Confusing Rejections?
Hi there,
Just wanted to share some thoughts (and vent a bit) after the IJCAI 2025 notifications dropped last night.
A Record-High Acceptance RateThis year, IJCAI hit a 19.3% acceptance rate, which is the highest in the past five years. Here’s the trend based on the numbers reported:
Year Submissions Acceptances Acceptance Rate IJCAI 2021 4204 587 13.9% IJCAI 2022 4535 679 14.9% IJCAI 2023 4566 643 14.1% IJCAI 2024 5651 791 14.0% IJCAI 2025 5404 1042 19.3% So yeah, on paper, things are looking good for IJCAI this year in terms of openness. But…
🤯 Confusing Reviewer Decisions
Many people were left scratching their heads when they saw meta-reviews marked "Accept" but still got rejected in the final decision.
Take this example that’s been circulating online:
“What is your suggested final decision? Accept. Note that your proposed decision may be overruled by the AC or Program Chair…”
Well, overruled it was. The paper was ultimately rejected.
This has led to a wave of confusion. Some say it’s not uncommon: AC accepts, but PCs or chairs veto it.
Others argue that most top-tier conferences do this internally but don’t show the contradictory info so explicitly. IJCAI just happens to be more "transparent" about the mess.
Either way, the inconsistency between meta-reviews and actual decisions has sparked a lot of debate. Some are calling it disorganized. Others think this transparency at least gives us insight into how chaotic things can be behind the scenes.
Shoutout to everyone who did get accepted — huge congrats to you!
Anyone else experience weird decision conflicts this year? Would love to hear how your reviews went.
