Why You Should Stop Using AI to Write Your Posts

Bryan Cantrill argues that the surge of AI-generated content on LinkedIn is ruining the platform's authenticity and annoying readers. While LLMs are useful for editing, their distinctive and grating writing style makes it obvious when a post isn't human-authored. He urges professionals to stop using AI for writing and return to using their own unique voices to maintain trust.

Key Points

• LinkedIn has become a primary social network by default because it remains stable while others implode.
• LLM-generated content is easily recognizable due to specific stylistic 'tells' like single-sentence paragraphs and repetitive phrasing.
• Using AI for authorship erodes trust and makes readers question the authenticity of the person's perspective.
• LLMs are valuable for editing and brainstorming but fail as a replacement for a human writer's unique voice.
• Authentic human writing is necessary to ensure a message is taken seriously and keeps the audience engaged.

Sentiment

The overall sentiment is supportive of the article's warning, with a strong preference for human-authored voice and skepticism toward machine-authored personal prose. The community is not broadly opposed to using LLMs as tools, but it is hostile to unedited, generic AI writing presented as personal expression. Opposition is present and substantive, focused on accessibility, productivity, and judging writing by usefulness rather than provenance, but it is outweighed by concern over trust, authenticity, and intellectual laziness.

In Agreement

• Writing is a form of thinking, and outsourcing it to an LLM means skipping the intellectual work that clarifies ideas, exposes weak arguments, and develops understanding.
• Readers value human judgment and voice; obvious machine-authored prose makes them question whether there is any real person, experience, or conviction behind the text.
• Default LLM prose is often padded, formulaic, and unpleasant to read, so publishing it without substantial editing signals disregard for the reader's time.
• The friction of writing is useful because difficulty in expressing an idea often reveals that the idea is not yet understood well enough.
• AI is acceptable for narrow assistance such as proofreading, grammar fixes, brainstorming, or translation, but not for generating the substance, structure, or voice of a personal post.
• LLM-generated posts resemble a new form of blogspam: easy to produce, easy to spread, and frustrating because they fill real communication channels with generic filler.

Opposed

• The message should matter more than whether a human or LLM produced the wording, especially for routine or low-stakes writing.
• LLMs can help people who are not strong writers, non-native English speakers, or people who dislike writing communicate more clearly and confidently.
• Some writing is administrative overhead rather than meaningful thought, so delegating it to a tool can be a rational use of time.
• Claims that writing always equals thinking are too broad, because some human writing is copying, dictation, formula, or low-effort expression.
• With better prompting and editing, LLM output can be concise and useful, so the problem is poor use rather than AI assistance itself.
• Resistance to AI-written text may be a purist aesthetic preference that becomes less practical as the tools become common.