AI/ML Research (GenAI:LLMs/VLMs, Agents and Reasoning, MedAI, HAI-Safety) 📝 Explorer ⚙️ Seeking Research Intern/Fellowship Opportunities ⏳Views are mine
England, United Kingdom
Joined November 2023
- Tweets588
- Following2.6K
- Followers496
- Likes2.4K
Azmine Wasi ✈️ NeurIPS'26 retweeted
People might overvalue papers from the pre-LLM era.
A garbage is a garbage. It doesn’t become more useful because it is fully generated by humans.
Similarly, a good paper is a good paper. It doesn’t devalue because AI is used to improve its quality.
Azmine Wasi ✈️ NeurIPS'26 retweeted
I don't think we should attribute 60k papers submitted to ICLR to AI. It's a natural evolution of bad incentives, and AI is the tool that fulfills them with minimal effort.
When administration and funding agencies count the number of published papers, everybody figures out the "minimal publishable unit" and splits one solid paper into four crappy ones.
When top AI professors demand NeurIPS publication from applying master's students, each student tries to win the publication lottery with something small and insignificant.
Many papers I read in the past year could have been a tweet with an attached GitHub repository of experiments. We got 15+ pages of write-ups about one-line changes (think something like DyT). Or a brutal mumbo-jumbo mixture of techniques to get improvement in the 3rd decimal place (in many disciplines this is called p-hacking, but we kind of ignore that).
Don't blame AI, blame the system.
Azmine Wasi ✈️ NeurIPS'26 retweeted
Something a few PhD students discussed with me already: since no one cares about conference prestige anymore except institutional bodies, they’re fully optimising their paper for ChatGPT reviews. They fine near 100% agreement between their final score and sufficient GPT reviews.
Azmine Wasi ✈️ NeurIPS'26 retweeted
It seems insane, completely indefensible to me, to say the code was never the hard part. Learning how to code at a high level would take you all the way to senior software engineer/architect. The only people saying this simply have no idea how things really worked in software
Azmine Wasi ✈️ NeurIPS'26 retweeted
Why waste the effort of review, rebuttal, discussion just to reject papers based on information that was available before the reviewers were even assigned??
#NeurIPS
Azmine Wasi ✈️ NeurIPS'26 retweeted
Welcome to Neur-RIPs 🪦
Rest In Peace to all our rebuttals, being cremated straight into the OpenReview void.
#NeurIPS2026
Azmine Wasi ✈️ NeurIPS'26 retweeted
I think initiatives like @icmlconf offering free registration to gold reviewers are a great example.
Since @iclr_conf already introduced a 20-paper submission quota per person this year, why not further adjust quotas based on review quality and responsibility? The more reliable you are as a reviewer, the more submission quota you should earn.
Ultimately, it would be great to move toward a real OpenReview system where everyone’s review credits and contributions are transparent and visible.
AI conference peer-reviews are completely broken. For NeurIPS 2026, my batch has very low reviewers responses rate (probably less than 5%), given that I have sent reminders twice. Maybe folks are all busy with their own rebuttals; Also for the @emnlpmeeting ARR submissions, the meta reviewer made factual mistakes and asking tons of ablation studies, however our submission is purely a dataset and benchmark paper, and we made it explicitly clear in all places.
If people don't really want to spend time even understand the paper you are reviewing, then it's probably good to just cancel the conference. What's the point of going through the review processes for months with people not even care what the paper is about? The field is completely broken. I have seen it consistently across multiple conferences.
Azmine Wasi ✈️ NeurIPS'26 retweeted
Reviewing my AC stacks for conferences this year, I’m more convinced that “human vs. AI review” is the wrong debate. The real divide is responsible vs. irresponsible reviewing. A careless human review can be far more harmful than a thoughtful AI-assisted one. How should conferences enforce accountability?
AI conference peer-reviews are completely broken. For NeurIPS 2026, my batch has very low reviewers responses rate (probably less than 5%), given that I have sent reminders twice. Maybe folks are all busy with their own rebuttals; Also for the @emnlpmeeting ARR submissions, the meta reviewer made factual mistakes and asking tons of ablation studies, however our submission is purely a dataset and benchmark paper, and we made it explicitly clear in all places.
If people don't really want to spend time even understand the paper you are reviewing, then it's probably good to just cancel the conference. What's the point of going through the review processes for months with people not even care what the paper is about? The field is completely broken. I have seen it consistently across multiple conferences.
Azmine Wasi ✈️ NeurIPS'26 retweeted
NeurIPS should not let authors reply to reviewers’ responses to their own rebuttal if they haven’t responded to the rebuttals of papers they reviewed. The authors of every paper I reviewed replied within an hour after I responded to their rebuttal, yet not a single reviewer has responded to mine.
Azmine Wasi ✈️ NeurIPS'26 retweeted
AI conference peer-reviews are completely broken. For NeurIPS 2026, my batch has very low reviewers responses rate (probably less than 5%), given that I have sent reminders twice. Maybe folks are all busy with their own rebuttals; Also for the @emnlpmeeting ARR submissions, the meta reviewer made factual mistakes and asking tons of ablation studies, however our submission is purely a dataset and benchmark paper, and we made it explicitly clear in all places.
If people don't really want to spend time even understand the paper you are reviewing, then it's probably good to just cancel the conference. What's the point of going through the review processes for months with people not even care what the paper is about? The field is completely broken. I have seen it consistently across multiple conferences.
Azmine Wasi ✈️ NeurIPS'26 retweeted
Peer review conference systems are weird because you have to convince "average people" that your completely wild thing is meaningful, which doesnt work, so you have to make most generic type shyt as much as possible. Hence everyone works on whats safe, which is boring generic topic that actually dont contribute to anything.
too many academics working on meaningless post training tweaks, harness / prompting bullshit, architecture tweaks, and muon variants.
Azmine Wasi ✈️ NeurIPS'26 retweeted
No one:
Today's NeurIPS reviewers if it was 2015: "ResNet is incremental"
Azmine Wasi ✈️ NeurIPS'26 retweeted
I deeply resent that AI has forced me to eliminate em dashes from my writing for fear of signaling slop
Azmine Wasi ✈️ NeurIPS'26 retweeted
💯! Funding, success, and most life opportunities are mostly a game of roulette anyway. So whatever happens you can always look back and say "at least I had fun!".
Azmine Wasi ✈️ NeurIPS'26 retweeted
To @AnthropicAI: don't be evil. You are acting like the biggest villain of the AI world.
