53% of TMLR submissions desk-rejected: 3 of 7 authors couldn't answer basic questions about their papers
Dr_Atoosa · x · 2026-09-18
TLMR editor-in-chief Nihar Shah says researchers submitting AI-assisted papers they don't understand is a "serious problem." Desk-rejection rates at the journal have climbed from 6% in 2023 to 53% now. Of 7 desk-rejected authors he interviewed, 3 couldn't answer basic questions, 3 struggled with technical details, and 1 — who answered everything — identified a major flaw in his own paper. Shah's team proposes using AI systems to question authors about their understanding of their own submissions, arguing authors bear ultimate responsibility for their papers' integrity.
Related event: TMLR Desk Rejects Surge to 53% as Authors Fail to Grasp Their Own AI Papers(4 posts)→
More from AGI Musings
- Debate: superintelligence could exploit human vision even if predators never did — basedjensen · 2026-09-18
- banteg amplifies critique of AI safety culture: longtermist moral math as 'mass delusion' — banteg · 2026-09-18
- New paper analyzes how AI disrupts universities and offers 10 survival proposals — LuizaJarovsky · 2026-09-18
- New paper "The University after AI" proposes ten ways to restructure universities — LuizaJarovsky · 2026-09-18
- banteg: longtermist moral math 'doesn't survive an ordinary discount rate' — banteg · 2026-09-18
- Ben Recht: what compressed sensing teaches us about AI's 'conquest of mathematics' — beenwrekt · 2026-09-18