85% of LLM-Generated Writing Contains at Least One Major Smell
A staggering 85% of writing generated by Large Language Models (LLMs) contains at least one major smell, according to recent observations. This phenomenon is particularly concerning for individuals who rely on LLMs to polish and enhance their writing, such as bloggers and content creators. The LLM-generated writing may feel significantly better than human-written content, but it often lacks the nuance and depth that a human writer can provide.
Overview
LLM smells refer to the subtle yet noticeable issues that can arise when using LLMs to generate writing. These smells can range from minor annoyances, such as inconsistent tone or style, to major problems, like factual inaccuracies or plagiarism. Despite the impressive capabilities of LLMs, they are not perfect and can produce writing that feels formulaic, lackluster, or even misleading. For instance, a math blog that uses LLMs to generate content may find that the writing is polished but lacks the depth and insight that a human writer can provide.Why It Matters
The prevalence of LLM smells matters because it can undermine the credibility and trustworthiness of the content. When readers detect even a hint of artificiality or manipulation in the writing, they may become skeptical and lose interest. Furthermore, LLM smells can also have serious consequences, such as spreading misinformation or perpetuating biases. As LLMs become increasingly popular, it is crucial to address these issues and develop strategies for mitigating them. By doing so, writers and content creators can harness the power of LLMs while maintaining the integrity and quality of their work.How to Start
To start addressing LLM smells, it is essential to understand the underlying causes. One major factor is the lack of human judgment and oversight in the writing process. LLMs rely on complex algorithms and statistical models to generate text, but they often lack the nuance and critical thinking that a human writer can provide. To overcome this limitation, writers can use LLMs as a tool to augment their writing, rather than relying solely on the generated content. By reviewing and editing the output, writers can inject their own voice, perspective, and expertise into the writing.Common Pitfalls
One common pitfall when using LLMs is over-reliance on the generated content. This can lead to a lack of originality and creativity, as well as a failure to fact-check and verify the accuracy of the information. Another pitfall is neglecting to consider the context and audience for the writing. LLMs may produce content that is generic or one-size-fits-all, rather than tailored to the specific needs and interests of the reader. To avoid these pitfalls, writers should use LLMs judiciously and with a critical eye, recognizing both the benefits and limitations of these tools.Recommendations
To minimize LLM smells and produce high-quality writing, consider the following strategies:- Use grammar and spell check software to review and edit the generated content, ensuring that it is free from errors and inconsistencies.
- Invest in content optimization tools that can help refine the tone, style, and structure of the writing.
- Utilize fact-checking and research software to verify the accuracy and credibility of the information.
- Consider language translation software to expand the reach and accessibility of the content.
- Explore writing assistants and productivity tools that can help streamline the writing process and improve overall quality.
By implementing these strategies and being mindful of the potential LLM smells, writers and content creators can leverage the power of LLMs while maintaining the integrity and excellence of their work. The next step is to start experimenting with these approaches and finding the right balance between human judgment and LLM-generated content. Begin by reviewing your current writing workflow and identifying areas where LLMs can be used to augment and enhance your writing.
What People Are Saying About Various Llm Smells
- Various LLM smells 28 May, 2026 Late last year I started writing a math blog and decided to use LLMs to polish/enhance my writing.
Sources & Context
Reporting and discussion this guide draws on:
- Various LLM Smells — Hacker News
All sources are linked. Excerpts are quoted under fair use to give you context before clicking through.
Recommended Tools
Some picks below for Various Llm Smells. Links use affiliate codes when available — your purchase price stays the same.
- Various Llm Smells subscription — Compare prices
- Various Llm Smells starter kit — Compare prices
- best Various Llm Smells tools — Compare prices
- Various Llm Smells for beginners — Compare prices
Get Weekly Picks Like This
Subscribe to the free newsletter — one curated email per week on Various Llm Smells and related niches.
Want the Deep-Dive Pack?
The Various Llm Smells Starter Pack condenses everything in this guide plus printable checklists, templates, and a 30-day plan. See pack →
Sponsored by your brand? Get in front of this audience.