Can someone explain why LLM's write like this when most humans don't?
I suppose it might be because humans that use LLMs write like this.
LLM Writing Tropes.md
71–80 of 211 posts
Re: LLM Writing Tropes.md
#72Another popular one is ending headlines with a remark/alternative in parentheses. Especially "(why this matters)".
More generally, it's interesting that many different LLMs have differences in their favorite tropes but converge on broadly similar patterns. Of course ChatGPT and its default persona (you can choose others in the settings, but most people don't do that) is overrepresented in these examples. For example, the article doesn't mention the casual/based tone of Grok that often feels somewhat forced.
Re: LLM Writing Tropes.md
#73Can someone explain why LLM's write like this when most humans don't?
I asked ChatGPT about that and it gave a nicely reasoned explanation on what AI produces compared to humans.
But that being said, the problem I think is that people treat the output from LLMs as final.
It should be treated more as idea generation or early draft to get over the “staring at a blank page” and get the creative juices flowing and creating your own content.
Having purely AI generated content and eventually feeding the algorithms and soon enough every sounds the same (already does in a lot of places).
Re: LLM Writing Tropes.md
#74No mention of Claude/ChatGPT's favourite new word genuine and friends? They also like using real and honest when giving advice. Far as I can tell this is a new-ish change. > Honestly? We should address X first. It's a genuine issue and we've found a real bug here. Honorable mention: "no ". I guess this helps reassure adherence to the prompt? I see that one all the time in vibe coded PRs.
I found a new one in claude recently with "Fair enough, ..."
Re: LLM Writing Tropes.md
#75Re: LLM Writing Tropes.md
#76What do people expect? You use an LLM, don't tell it your preferred writing style and get annoyed when it falls back to defaults.
All those tropes have their place in certain contexts. AI overusing them is because they have no memory across all they've written.
Each conversion is a new chat so it's like "I haven't used delve in a while, think I'll roll out that bad boy"
And then you try to fix this by telling it what not to do which doesn't work very well, so...
Re: LLM Writing Tropes.md
#77Earlier quoted context omitted.
Weird to care about a harmless construction along with punctuation.
Construction paired with punctuation is literally the entire point of written communication.
No it's just the medium. The point is to communicate.
You can test this quite easily, by checking and hopefully realizing that you in fact can understand written documents with syntax errors, emails with typos and road signs with improper casing or sentence construction.
Re: LLM Writing Tropes.md
#78This changes everything.
You're absolutely right!
Re: LLM Writing Tropes.md
#79A subtle tell for generated text is just how damn flat it is to read. Not that technical documentation require some form of grand prose, but how unspecific the text can truly get. Reading a high school persuasive essay can have more detail, and those are often just written for a grade.
I can understand someone needing help with writing but getting an agent to do the job for you feels like a personal defeat.
Re: LLM Writing Tropes.md
#80Honestly, you need a tailored one of these for each of the major LLM model/version pairs. Claude and Gemini don't exhibit all of the same tropes in the same severities as OpenAI's GPT series, and within each of those, each revision sometimes exhibits substantial variance from the stylistic propensities of its immediate predecessor.