An Vo

phd student @UMichCSE | prev: @mbzuai, @kaist | llms/vlms, trustworthy ai

[An Vo @an_vo12]

Life update: I've joined the University of Michigan (@UMichCSE) as a PhD student with @radamihalcea.

At this pivotal moment for AI, I hope my PhD research will contribute to a better understanding of AI and make it safer and more trustworthy.

[An Vo @an_vo12]

This work was done during my wonderful time @mbzuai!

Huge thanks to my special co-first author @TuongVy115549 for all the hard work on this project.

Grateful to Daeyoung Kim, @thamar_solorio , and @anh_ng8 for advising this project, and to @knnguyen2511 and @evllcv for their amazing contributions! 🙏

[An Vo @an_vo12]

Heading to #EMNLP2026 in Budapest 🇭🇺 to present our paper "Deep and shallow biases in language models"!

Not all LLM biases are equal.

🎲 Ask an LLM for a random number → 42. Reword it → now it says 3.
🦋 Ask for a random butterfly → Monarch. Reword it → still Monarch.

Why do some biases survive rewording and others don't?

🎬 Watch the video below and learn more at https://t.co/BGSvA5kyFK

[patrick @patrickamadeus_]

dropping something soon 👀

[Phillip Isola @phillip_isola]

Students sometimes ask me if it still makes sense, in this accelerating age, to pursue a PhD in AI.

Perhaps counterintuitively, I think it's a great time to do so.

I wrote up some thoughts on this here: https://t.co/RRqGR0qi1R

[An Vo @an_vo12]

@NaihaoDeng yeah i'm also curious about this too! there's nothing wrong with using AI to polish text here, especially when many non-native speakers (including me) are using AI every day as an effective tool to get feedback and improve their writing skills.

[An Vo @an_vo12]

Is it game over for the peer review system? If not, how much longer can it keep up?

Something needs to change. I wonder if it’s time to cap submissions at 5 per author!?

[Jacob X. Li @jacobli99]

60k+ ICLR 2027 submissions is more than all prior ICLRs combined

[Logan Bolton @septisum]

Reran with 3 open-source models.

Kimi looks excellent, Deepseek is not very good and GLM Flash is pretty nice.

Opus is better than all of them by far though. Other people were saying that Anthropic might be stealth routing requests to the new Opus that is dropping soon?

[Transactions on Machine Learning Research @TmlrOrg]

TMLR has faced a deluge of submissions, necessitating stricter desk rejection policies due to limited reviewer capacity

Co-EiC Nihar Shah reached out to authors of 10 papers slated for desk reject. Could they answer questions about their *own* submission?
https://t.co/vhX5w6gcx2

[MichiganAI @michigan_AI]

Welcome aboard @an_vo12! 🎉
We’re excited to have you join our community and look forward to all that you’ll contribute and accomplish.

[An Vo @an_vo12]

@patrickamadeus_ what if they release 6.1 tomorrow? 😂

[An Vo @an_vo12]

@septisum Haha I used to open my papers like that a few years ago, but these days I prefer to get straight to the point to save some space :D

[ACL 2027 @aclmeeting]

ACL Sustainable Reviewing Policy: We are introducing changes in the @ReviewAcl reviewing and submissions. The changes will cap authors and introduce changes in the reviewing to keep our community sustainable. #NLProc

[Joan Nwatu @_agirlyengineer]

🌎 A dataset says it covers Spanish. But whose Spanish?
Spanish is spoken across Spain and much of Latin America, from Mexico to Argentina and beyond. Yet an NLP dataset labeled simply “Spanish” may contain data representing only a small subset of those countries.

[An Vo @an_vo12]

@cai_xukun33535 @thamar_solorio @mbzuai It depends on how competitive each program is. Prior research experience would help (eg, a first-author paper, projects) but you don't need to have everything.

Strong fundamentals and communication skills help too, especially in interviews. Hope this helps, and good luck!

[An Vo @an_vo12]

Life update: I've joined the University of Michigan (@UMichCSE) as a PhD student with @radamihalcea.

At this pivotal moment for AI, I hope my PhD research will contribute to a better understanding of AI and make it safer and more trustworthy.

[An Vo @an_vo12]

@cai_xukun33535 @thamar_solorio @mbzuai I can also add some perspective from my experience. A research-based MS is another great option!

I did mine at KAIST with full funding & good resources, and it was a great chance to learn how to do research. Similar opportunities exist elsewhere in Asia too (e.g., MBZUAI?).

[An Vo @an_vo12]

@prathoshap Had a similar experience with my NeurIPS review batch. Not sure what a good screening process would look like but a submission cap might help. Maybe cap submissions at 5 per author and 1 for first-time authors (like ICLR did).

[An Vo @an_vo12]

@AnganaBorah2 @aaclmeeting @ZhijingJin @radamihalcea congrats Angana!

[An Vo @an_vo12]

@OfirPress i'd love to see more benchmarks that are hard for AI but easy enough for a kid. Feels like those could tell us a lot about what AI is still missing

[An Vo @an_vo12]

@NaihaoDeng @UMichCSE @radamihalcea I've heard a lot about. Excited to see you next week Naihao :)

[An Vo @an_vo12]

@patrickamadeus_ @UMichCSE @radamihalcea looking forward to seeing your next work Patrick