LLM-written conference reviews are making 1-page rebuttals look outdated
CSProfKGD · x · 2026-07-28
The author says conference rebuttal policies may need to change as more reviews are being written by LLMs.
- They argue that a single-page PDF rebuttal worked when reviews were human-written.
- But when an LLM can generate 8–15 negative points, authors are forced to squeeze responses to both human and machine reviews into one page, which can lower rebuttal quality.
- The post suggests conferences should reconsider rebuttal format now that fully LLM-generated reviews are hard to prevent.
- The author adds that, despite the extra work, there was one upside to the older rebuttal style, but the excerpt cuts off before finishing that point.
More from Research
- A visual PCA explainer walks through centering, eigenvectors, and variance capture — mdancho84 · 2026-07-28
- Why world models are being framed as the next big step in AI — JackFisherBooks · 2026-07-28
- Open-source benchmark tracks coding-agent token usage without leaking prompts — bestofdesp · 2026-07-28
- Researcher argues causal claims can still be drawn from observational data — yanaiela · 2026-07-28
- A stronger model can still hurt your AI product, so test upgrades on production tasks first — max_gladysh · 2026-07-28
- RepLLM uses four agents to reproduce network-system papers in about two hours — 机器之心 · 2026-07-28