Hacker News Reader

Comment by chrisjj | Hacker News Reader

I just asked ChatGPT to put borders around some simple rectangular HMML elements. The result has gaps on some and not on others. Some have gaps on some sides and not others. One has a gap when stationary but not when dragged. When I request it be fixed, the bot says it is fixed - and it isn't.

You may call that variance. I call that unstable.

A human taking such a task generally does a lot better.

View on HN

PS Anyone else relying on ChatGPT's online preview should know it secretly places an invisible overlay to intercept input. Goodnees knows why, but thus interferes with touch input to your HTML, leading to use time wasted chasing non-existent bugs.

yes you're bad at using it. Your mistake is presuming everyone else has that problem.

Try considering the input a search query and the output a list of results of which one is arbitrarily chosen. If you get your query so specific that the amount of possible results are small enough that all of them are good, then it works. Big longer starting context, hand-written, use the LLM to improve it by criticising it (but keep it mostly hand written), aim to do more in a corresponding YOLO session.

> If you get your query so specific that the amount of possible results are small enough that all of them are good, then it works.

I guess I'll try giving it the finished program.

give it a full spec of what you want. Full waterfall. You can then take it into an agentic session and get a lot out of it.

Reply on news.ycombinator.com