Ask HN: Almost unusable code generated by ChatGPT recently

6 points by illegalmemory 2y ago ↗ HN

Lately, I've noticed issues with ChatGPT generating code that is nearly unusable, often leaving massive gaps in the code and prompting me to fill them up. Have you experienced similar outputs?

5 comments

[ 2.7 ms ] story [ 26.0 ms ] thread

verdverm 2y ago ↗

There have been several recent HN posts to the same effect. Here's a recent one I participated in: https://news.ycombinator.com/item?id=38602025

It's not limited to ChatGPT either and seems more pervasive. Not sure what's happening

illegalmemory 2y ago ↗

Looks like it! Even code quality has dropped 'significantly.' I used it once in a while to clean up the code after writing a POC. It used to do decent work before, but now I almost always can't use the code generated at first shot.

ipaddr 2y ago ↗

This is never reflected in the benchmarks and everyone uses them as imperial universal truths. ChatGPT once reached a certain quality but is scaling down on first requests to save compute power.

j4hdufd8 2y ago ↗

learn to code

runjake 2y ago ↗

Data point: it works well here (on a Plus subscription using GPT-4) with Python and JavaScript and Go.

I do use highly-tailored starting prompts saved in a GPTs.

It's definitely gotten a lot slower.