I Can’t Believe It’s Not Better Initiative

(6/n) Yutaro Yamada (Sakana AI) will present “The AI Scientist-v2” — an agentic system that wrote & submitted papers to our workshop. What does this mean for peer review, novelty, and research norms?

(5/n) Roberta Raileanu (Meta) will talk about using LLMs to automate scientific discovery—how far are we really? She’ll introduce MLGym & show why real-world AI research is still far from solved.

(4/n) John Kalantari (YRIKKA / Univ. of Minnesota) will tackle the myth of generalizability in DL—why context isn’t a footnote, but the key to real-world impact. Especially in high-stakes fields like healthcare.

(3/n) Otilia Stretcu (Google Research) will unpack why classification still breaks in the real world—and how LLMs might help (or not). From safety to niche data, it’s messier than the benchmarks suggest.

(2/n) Nick Haber (Stanford) will explore where reasoning-capable LLMs still fall short — and what that means for learning, education, and human-AI collaboration.

🔥 Speaker spotlight for I Can’t Believe It’s Not Better #ICLR2025!
(1/n) Sunayana Sitaram (Microsoft Research) will dive into how LLMs fail in non-dominant languages—mistranslations, bias, and cultural blind spots. What does it mean for inclusion in AI?

ICLR 2025 starts next week!
Our workshop “I Can’t Believe It’s Not Better” is on April 28 (Singapore time) — join us to talk real-world challenges in applying foundation model and deep learning across fields.

ICLR Workshop 2025 - Call for Papers We invite researchers and industry professionals to submit their papers on negative results, failed experiments, and unexpected challenges encountered in applying deep learning to real-world problems ...

Our #ICLR2025 @iclr-conf.bsky.social workshop is looking for submissions on unexpected outcomes and hard-earned lessons in real life #DeepLearning
Submission Deadline: 03 February 2025
Workshop Dates: 27 or 28 April 2025
Location: Singapore
More info: shorturl.at/OpQns

ICLR Workshop 2025 - Call for Papers We invite researchers and industry professionals to submit their papers on negative results, failed experiments, and unexpected challenges encountered in applying deep learning to real-world problems ...

Please spread the word! Our “I Can’t Believe It’s Not Better” #ICLR2025 workshop @iclr-conf.bsky.social is now accepting submissions on unexpected results in applied #DeepLearning. Share your tough lessons and unexpected outcomes by Feb 3, 2025. More info: shorturl.at/tD1ju

The call for papers is out for our “I Can’t Believe It’s Not Better” #ICLR2025 workshop @iclr-conf.bsky.social . Don’t miss this chance to highlight challenges in applying deep learning & foundation models in real life.

Please help spread the word. Our flyer:

ICLR Workshop 2025 - Call for Papers We invite researchers and industry professionals to submit their papers on negative results, failed experiments, and unexpected challenges encountered in applying deep learning to real-world problems ...

Attention ML researchers! We invite you to submit your work to the “I Can’t Believe It’s Not Better” workshop at #ICLR2025. Let’s shine a light on those unexpected outcomes, elusive improvements, and the tough lessons learned in applied #DeepLearning. Deadline is Feb 3, 2025. More details: