HCI+ML researcher and professor of Computer Science at @uni_lu. Previously with @AaltoUniversity and @UPV. Also co-founder of @ScilingAI in 2014.
Luxembourg
Joined February 2011
- Tweets3.2K
- Following474
- Followers803
- Likes7.4K
Pinned Tweet
🧵 Some reasons to come to Luxembourg for research.
ALT Image credit: Wolfgang Staudt - https://www.flickr.com/people/53074617@N00
Luis A. Leiva retweeted
At this point, ICLR and OpenReview feel more like data collection and labeling platforms. I did 3 HITs, 2 labeled as slop and 1 full review. AISTATS pile is looking much better. Less popular venues are still spared from the DDoS.
Anyways, lots of cool benchmarks/datasets to come out once the ICLR data is public.
OpenReview and arXiv can use ICLR data to train a model to classify slop vs. no-slop and use it to desk-reject.
Luis A. Leiva retweeted
when the cybersecurity team finds out there are actual vulnerabilities and they have to do their jobs instead of trying to trick someone into clicking a phishing email
Luis A. Leiva retweeted
Craziest arXiv paper this month:
"The author has not independently checked the analyses and does not have the expertise to verify every technical detail"
arxiv.org/pdf/2610.28575
Meanwhile, arXiv blocked 島袋隼士's paper @bukuro8810
Luis A. Leiva retweeted
Some thoughts on reviewing: spylab.ai/blog/reviewing/
First for @satml_conf where we trialed an early reject phase which cut submissions by ~40%.
Then more generally about how not everything deserves peer-review, and why we should reward writing things that people actually read.
Luis A. Leiva retweeted
SQL Injection was publicly documented 28 years ago. Parameterized prepared statements take zero extra lines of code in every language on Earth.
And yet right now, at this exact second, somebody is concatenating raw user input straight into a SELECT * FROM users WHERE string in production.
Luis A. Leiva retweeted
Try opening a Microsoft Word document from 1992, a Flash project from 2006, or a proprietary database dump from 2010. Half of it is unreadable binary rot.
Now open a .c source file or a plain .txt file written on a UNIX PDP-11 in 1973. It opens in 0.001 seconds on every computer on Earth.
Plain text is the only truly immortal data format humanity has ever invented.
Luis A. Leiva retweeted
New in Science: we randomized phone-free classrooms for 16,000+ students across 10 institutions in Odisha, India. Students deposited phones in a box at the start of each class, for a full semester.
Luis A. Leiva retweeted
I'm not super up to date with the latest slang and terminology folks are using these days, but I believe these are called "schools"
Luis A. Leiva retweeted
going to tell my niece and nephew that the bottom half of this is simon and garfunkel and that the top half is AI.
(if everyone is going to keep rewriting history, then we might as well have a little fun with it.)
Luis A. Leiva retweeted
Both are true. Research has accelerated and the friction to make an idea into a paper has gone down 10x or more. At the same time, the cost of writing a slop paper has also gone down, and in general good ideas are more likely to be painstakingly turned into papers than bad ideas. In the past, someone would think "why would I bother write this not so great paper", while now they might think "oh well, let me have Codex turn this into a paper in a couple of days, why not".
Luis A. Leiva retweeted
(1/n)
The number of research papers has gone wild. All generated by AI?
I pulled stats from CVPR/ICCV/ECCV, NeurIPS/ICML/ICLR, SIGGRAPH/SA/ToG since 2006: 119k papers + 460k arXiv preprints
Accepted/yr: 968 (2006) → 19,003 (2025) → 18,976 (2026 so far)
Quite surprising ->🧵
Luis A. Leiva retweeted
To me the biggest risk of AI is brain rot.
I have more and more situations like this. Some coworker shares a report on something they did. I read, and find something fishy. I ask about and get no answer. I ask again and I'm told: the agent said that. And no follow up. It's clear form the interaction that they did not read the report they shared. Their brain is not used as it should be.
I am not against AI at all (fortunate for something working at NVIDIA). I use AI a lot.
But I use it as a tool. I own the result and whenever I see something fishy I ask about its output I force a change if I am not convinced.
I wish there was a way to force people to read the AI output they share with others.
Luis A. Leiva retweeted
1/ ICLR 2027 got 60K+ submissions. If we want the conference format to survive (I do), we have to rethink the whole thing. Some suggestions:
