LLMs changed writing. But our submission processes haven’t caught up.

We are entering a strange communication loop. One person uses an LLM to generate a long proposal, report, or application document, and the receiver then uses another LLM to summarize it into bullet points.

We are effectively using AI to expand text, only to use AI again to compress it.

A lot of submission processes, grant applications, RFPs, and evaluation forms were designed for a pre-LLM era, when producing long, polished documents required significant human effort. Length itself was often interpreted as effort, seriousness, or competence. Today, everyone has access to tools that can generate pages of convincing text in seconds, and this changes the value of written communication.

If the real goal is to communicate ideas, requirements, risks, timelines, or proposals, then we should optimize for clarity and essence instead of document length. Personally, when I receive long documents nowadays, I often feel they were not truly written for humans to read. In a fast-moving world, concise communication matters more than ever. Bullet points, summaries, key decisions, tradeoffs, and actionable information are usually enough.

Otherwise, the workflow becomes absurd: use an LLM to generate 20 pages, use another LLM to summarize the 20 pages, then read the summary.

We should adapt our systems and expectations to the reality of the tools we now have. Instead of requiring maximum text, we should require maximum clarity.