Live data from Hacker News

Ask HN: Are you also using ChatGPT a lot every day for code related work?

news.ycombinator.com

1–4 of 4 posts

Ask HN: Are you also using ChatGPT a lot every day for code related work?

#1
I find my self sending at least 20 to 50 messages to ChatGPT every day. I would probably need to check 2-5 links on Google, and combine information to solve the same.

Now I feel I solve more than 50x as much on open source projects. Faster feedback. Can iterate on my phone on a walk, and come back to push code.

The best part is that I feel I have a lot more energy and can do more of the fun part; solving problems.

The only “bug” I find is that I often have to ask ChatGPT to repeat the last part of the code, to get it to print everything.

Anyone else that is doing the same? And have some other tricks to speed it the development process?

Re: Ask HN: Are you also using ChatGPT a lot every day for code related work?

#2
I’ve been experimenting with both ChatGPT and Copilot. I find that Copilot is much more accurate and context-aware, but I can’t use it everywhere—e.g. when working in a shared Colab notebook. Conversely, I find that ChatGPT is better if I’m brainstorming some ideas and want to get a high-level sense of different ways to execute them in code.

This said, ChatGPT has a tendency to get things about 95% right, so you have to be very careful. For example, I was playing with it yesterday and it wrote a Python function with regex that didn’t do what I had asked, even though it said it did. I pointed out the error and it rewrote the regex but made it worse.

Oddly, when I started another thread and asked ChatGPT just for the regex outside of a Python function, it nailed it. So I do think prompt engineering can overcome some of the limitations I’ve seen to date.