Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

As great as ChatGPT is, the content it generates tends to follow some fairly repetitive patterns & styles, and it's going to be pretty easy to identify unless heavily edited.

However, I also suspect this technology will develop very rapidly: just look at how quickly the AI art scene has evolved since Stable Diffusion came along. Pretty crude at first, but the quality of stuff coming out of it is amazing now.



There are services to identify LLM output, like GTPZero and a classifier from OpenAI, but they aren't very reliable even at this early stage in the development of LLMs—lots of false positives and negatives. They'll likely be useless when the tell-tale signs they rely on are ironed out.

https://gptzero.me/

https://openai.com/blog/new-ai-classifier-for-indicating-ai-...


In my experience adding "in the style of a human writer" or "in the style of [author name]" is enough to break detection.


I feel like the biggest impact will be for authors who have an existing body of work... you can train the model on your writing and generate new text in your own style. What takes days could be done in hours.

I'm sure the same could be done for editors... train it on annotated text (original + corrections), use it to generate new edits.


(Copying my comment over from a different HN ChatGPT thread I commented on this morning - are all these ChatGPT threads getting a bit ... samey?)

As a writer, I'm much happier in the revision/editing stage than in the first draft stage. I have a half-finished novel that's been half-finished for the best part of 8 years now. I'm now thinking that maybe if I feed the draft into <ML-shiny-of-the-day> with suggestions for how I want the next section/chapter to develop, it could generate some draft copy for me. Possibly several different versions. After which I'm in my Happy Editing Place, shaping existing copy to something I like. Repeat and rinse!

The end result would be 100% my work as the ML algorithm/brains/whatever would be generating new copy based on my existing copy and my vision and direction for the novel. And I get the final say on the results. Win-Win!


Sure, but most authors aren't payed for their writing style, they are paid for what they say, for a unique insight, good research, or, if fictional, for a good story arc and suspense. I haven't seen purely AI written pieces that have that. And I am not sure they will with the current approach, which produces mostly bland and generic text. We would need an AI that does some actual creative thinking first.

I think (hope) that AI, in the interim, will provide actual authors ways to get inspired by its random nature. Like painters might start with an ink blob and then paint over it. Or let AI fill in some gaps, like description of people or places, etc.


Damn, I hadn't thought of that. I was thinking I could stay safe by just sticking with already known authors. But apparently even that won't work.


If you separate generating the content from stylizing it you can make it much more natural and targeted toward your readers. This generation of models can do a lot but does things best when you do one thing at a time.




Consider applying for YC's Summer 2026 batch! Applications are open till May 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: