Why LLMs Are Bad Writers But Good Editors
ivan_bezdomny · x · 2026-08-13
In an essay for The Atlantic, Jasmine Sun explores why large language models (LLMs) struggle with literary creation while excelling as editing tools.
The article notes that despite tech leaders predicting LLMs will solve climate change or colonize space, their capacity for art remains limited. The author identifies three core reasons:
- The Verifiability Problem: "Good writing" is highly subjective and hard to evaluate, making it difficult to optimize during model training.
- Business Priorities: Companies focus on post-training goals like safety and utility, which often interfere with literary quality.
- Lack of Grounding: Good writing requires a grounding in real-life experience, which AI inherently lacks.
However, the author argues that LLMs are highly effective as editors, and shares a practical Claude editor setup to assist human writers.
More from AGI Musings
- Opinion: LLMs Will Destroy Self-Referential Science, Verifiability Becomes Engineering — RexDouglass · 2026-08-13
- Who gets credit for AI-driven scientific discoveries? Not the chatbots — analisereal · 2026-08-13
- Acemoglu on Automation: Why It Won't Bring Shared Prosperity & Pro-Worker AI — danielrock · 2026-08-13
- Anthropic's Leaked Deck Predicts 2026 as the Last Window to Catch Up in AI — imjustnewatai · 2026-08-13
- MIT Professor Advocates for 'Intelligence Augmentation' Over Replacing Humans — round · 2026-08-13
- AI science has become about discovering frontier companies' secrets; open research is critical — rao2z · 2026-08-13