AI's top startups are barely publishing their research
TL;DR Highlight
Commenters debate why top AI startups rarely publish research, citing incentives and competition.
Who Should Read
A developer at an AI startup building on third-party models who must decide what research to publish.
Core Mechanics
- No article text was provided
- The article body was empty, so its own facts can't be verified
- The community reactions below come only from comments
Community Reactions
- Publishing costs more than it returns
- Several commenters said competitors copy published work and capture the revenue
- One said research becomes a pure cost center
- Papers serve as fundraising material
- One commenter spent 3 years trying tier-1 journals before settling on a preprint
- Their second startup publishes nothing for fear OpenAI and Anthropic will copy results
- Pushback on the article's framing
- A commenter said the paper used citations as a proxy for significance, with OpenAI ranked first in cumulative citations
- Others said their startups share work via blogs and open source instead of papers
- Concern about the field's health
- Some argued blog-style AI research spreads unsupported claims
- Others worried that if everyone withholds work, the whole sector slows down
How to Apply
- Write a one-page publishing policy
- Spend 30 minutes today listing what your team shares via papers, blogs or open source
- Mark anything that hurts you if copied as private
Terminology
Related Papers
The Judge Can Debias Itself: Awareness and Self-Debiasing for Reliable LLM-as-a-Judge
Warning an LLM judge about bias or letting it self-critique improved 20 of 21 model–bias combinations.
90% of US working hours are spent on tasks not covered well by AI benchmarks
Mapping 29 benchmarks to job tasks shows 90% of US working hours lack good coverage
Postgres 19 delayed after AI reviewers find more bugs than humans can fix
LLM code review surfaced so many bugs that Postgres 19 slipped to November and 8-10 features were reverted
Show HN: Gemini 4 Argon plays StarCraft
LLMs wrote StarCraft bots in one hour; GPT-6 Astra scored highest.
A Plausible Explanation Is Enough to Mislead Multimodal Judges
Adding plausible explanations to fixed answers cut MLLM judges' wrong-answer rejection by 12.7–13.8 points.
Neural networks are clean
Looking at neurons directly shows most have one main job, challenging superposition.