WingNews logo WingNews
top | item 43721301

Recursive LLM prompts

81 points| vlan121 | 10 months ago |github.com

19 comments

order

mertleee|10 months ago

"Foundational AI companies love this one trick"

It's part of why they love agents and tools like cursor -> turns a problem that could've been one prompt and a few hundred tokens into dozens of prompts and thousands of tokens ;)

danielbln|10 months ago

It's be nice if I could solve any problem by speccing it out in its entirety and then just implement. In reality, I have to iterate and course correct, as do agentic flows. You're right that the AI labs love it though, iterating like that is expensive.

ivape|10 months ago

The bigger picture goal here is to explore using prompts to generate new prompts

I see this as the same as a reasoning loop. This is the approach I use to quickly code up pseudo reasoning loops on local projects. Someone had asked in another thread "how can I get the LLM to generate a whole book", well, just like this. If it can keep prompting itself to ask "what would chapter N be?" until "THE END", then you get your book.

danielbln|10 months ago

The last commit is from April 2023, should this post maybe have a (2023) tag? Two years is eons in this space.

gwintrob|10 months ago

Crazy that OpenAI only launched o1 in September 2024. Some of these ideas have been swirling for a while but it feels like we're in a special moment where they're getting turned into products.

vlan121|10 months ago

I had a different title. It was somehow changed to the name of the repository.

jdnier|10 months ago

The author is Co-founder of Databricks, creator of K Prize, so an early adopter.

K0balt|10 months ago

This is kind of like a self generating agentic context.. cool. I think regular agents, especially adversarial agents, are easier to get focused on most types of problems though.

Still clever.

James_K|10 months ago

I feel that often getting LLMs to do things like mathematical problems or citation is much harder than simply writing software to achieve that same task.

mentalgear|10 months ago

Trying to save state in a non-deterministic system, not the best idea. Those things need to be externalised.

seeknotfind|10 months ago

Excellent fun. Now just to create a prompt to show iterated LLMs are turing complete.

ivape|10 months ago

Let's see Paul Allen's prompt.

mentalgear|10 months ago

Should definitely get a date tag.

vlan121|10 months ago

I was leaving this one out, seems like a gag when you read it :D