Hacker News new | past | comments | ask | show | jobs | submit login

Don't need an LLM for that (except maybe to get funding), a plain old classifier would work fine at a fraction of the training & inference costs.



Yeah, but why would you make that effort when all you'd need is the thinnest veneer over ChatGPT, and given the proposed pricing would leave plenty of margin?


Well you're not going to be able to charge them a lot of money without claiming LLM and you won't get funding, so gotta think about the big picture.


Don't worry, you can say something vague like "it's powered by LLM" which could reasonably mean that the LLM was used during the training phase of your own classifier.


Or that you asked ChatGPT how to write scikit-learn code.




Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4

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

Search: