FULL STORY

DeepMind researcher sparks debate on killing conference papers

A CMU professor and DeepMind researcher declared conference papers dead, urging researchers to publish via GitHub and blogs instead. The claim ignited wide debate over declining paper value and reproducible work on X gaining more attention than poster sessions.

2026-11-16 ~ 2026-11-17 · 3 episodes · 14 posts

Episode 1 · DeepMind researcher declares conference paper game dead, urges GitHub-first research output (2026-11-16, 9 posts)

Dimitris Papailiopoulos argues that publishing at conferences is now clearly the wrong channel for AI/ML research, and says he hasn't had any junior researcher submit to a conference in the past year. Peers largely agree but admit uncertainty about advising students, while some note offline networking remains crucial for students.

Episode 2 · Value of top AI conference papers collapses as costs rise, researchers say (2026-11-16, 3 posts)

Researchers including Dimitris Papailiopoulos and Yale's Amin Karbasi say the value of publishing at top AI conferences has plummeted while costs keep rising, with students still submitting mainly as a job-market signal and many turning to social media to disseminate research.

Episode 3 · MIT researcher: reproducible work on X beats conference posters for reach (2026-11-17, 2 posts)

MIT researcher Dimitris Papailiopoulos argued that immediately reproducible work posted on X can attract more attention than a random poster session at a top conference, noting that even low-follower young researchers often outperform NeurIPS/ICML posters on the platform.