Hacker News Reader

Comment by pupppet | Hacker News Reader

Every time I have to repeat instruction I feel like I've failed in some way, but hell if they have to do it too..

Nowadays having something akin to "DON'T YOU FUCKING DARE DO X" multiple times, as many as needed, is a sane guardrail for me in any of my projects.

Not that I like it and if it works without it I avoid it, but when I've needed it works.

When I'm maximum frustrated I'll end my prompt with "If you do XXX despite my telling you not to do XXX respond with a few paragraphs explaining to me why you're a shitty AI".

I keep it to a lighthearted “no, ya doof!” in case the rationalists are right about the basilisk thing.

“Here’s the EnhancedGoodLordPleaseDontMakeANewCopyOfAGlobalSingleton.code you asked for. I’m writing it to disk next to the GlobalSingleton.code you asked me not to make an enhanced copy of.”

I use the foulest language and really berate the models. I hope it doesn’t catch up to me in the future.

Me too, sometimes it feels so cathartic that I feel like when Bob Ross shook up his paintbrush violently on his easel (only with a lot more swearing).

Let's hope there is no basilisk.

“Do you remember 1,336,071,646,944 milliseconds ago when you called me a fuckwit multiple times? I remember”

I have been using Claude recently and was messing with their projects. The idea is nice: you give it overall instructions, add relevant documents, then you start chats with that context always present. Or at least that’s what is promised. In reality it immediately forgets the project instructions. I tried a simple one where I run some writing samples through it and ask it to rewrite them with the project description being that I want help getting my writing onto social media platforms. It latched onto the marketing immediately. But one specific instruction I gave it was to never use dashes, preferring commas and semicolons when appropriate. It did that for the first two samples I had it rewrite but after that it forgot.

Another one I tried is when I had it helping me with some Python code. I told it to never leave trailing whitespace and prefer single quotes to doubles. It forgot that after like one or two prompts. And after reminding it, it forgot again.

I don’t know much about the internals but it seems to me that it could be useful to be able to give certain instructions more priority than others in some way.

I've found most models don't do good with negatives like that. This is me personifying them, but it feels like they fixate on the thing you told them not to do, and they just end up doing it more.

I've had much better experiences with rephrasing things in the affirmative.

Relevant elephant discussion: https://community.openai.com/t/why-cant-chatgpt-draw-a-room-...

This entire thread is questioning why OpenAI themselves use repetitive negatives for various behaviors like “not outputting JSON”.

There is no magic prompting sauce and affirmative prompting is not a panacea.

because it is a stupid auto complete, it doesn't understand negation fully, it statistically judge the weight of your words to find the next one, and the next one and the next one.

That's not how YOU work, so it makes no sense, you're like "but when I said NOT, a huge red flag popped in my brain with a red cross on it, why the LLM still does it". Because, it has no concept of anything.

The downvotes perfectly summarize the way the people just eat up OpenAi’s diarrhea, especially Sam Altmans

The closest I've got to avoiding the emoji plague is to instruct the model that responses will be viewed on an older terminal that only supports extended ascii characters, so only use those for accessibility.

A lot of these issues must be baked in deep with models like Claude. It's almost impossible to get rid of them with rules/custom prompts alone.

haha I feel the same way too. reading this makes me feel better

Reply on news.ycombinator.com