With 60k ICLR abstracts, a researcher mourns LLM-generated figures homogenizing papers
mariyaivasileva · x · 2026-09-26
As she finalizes ICLR submissions, researcher Mariya Ivasileva reflects on losing the "old school" craft of paper writing. Her advisor insisted figures were the most important anchors — the first thing reviewers look at and often the only thing they remember — so she would iterate on visuals instead of drafting prose.
Key observations:
- LLMs can now one-shot graphics into "homogeneous oblivion," and fewer people ask "what would make this paper even better?"
- She still writes for both novice and expert readers, leads with the simplest representative example, and introduces notation only when meaningful
- Related-work sections are shrinking or demoted before the conclusion; references rarely reach back more than a year. With 60k abstract registrations this year, it's hard to imagine the work hasn't been done
She ends wondering whether research communication has fundamentally changed: maybe hardly anyone reads the paper, and everything interesting should just fit in the abstract.
More from AGI Musings
- Claim: China's AI progress is 'stunning', US has ~7 months to deploy open source plans — TinfoilTricorn · 2026-09-26
- Ex-OpenAI policy lead: friends flood him with AI worries, and he can't tell them it's overblown — Miles_Brundage · 2026-09-26
- "Fake AI Incidents Are a Bigger Threat Than Autonomous AI," Argues Srinivasan — robleclerc · 2026-09-26
- Dev Advice: AI Makes Platform-Heavy Side Projects the Best Career Bet — himanshustwts · 2026-09-26
- AI Safety Space Has Serious Sexism Problems, Predicts Public Assault Account Within 2 Years — ShakeelHashim · 2026-09-26
- US media's AI doom narrative shapes public fear; author sees no easy fix — Dr_Singularity · 2026-09-26