logoalt Hacker News

loloquwowndueo • today at 12:39 AM • 3 replies • view on HN

> The project code was written by GPT-5.6 Sol xhigh. It has not been verified by a human for correctness, efficiency, or speed; i.e. ymmv and use this at your personal risk.

In an age where everyone has access to AI, AI-written code has zero value. You should share the prompt or ask your agent to write a document other people’s agents can use to reimplement the code from scratch and/or customize to their specific needs.


Replies

areoform • today at 1:08 AM

    > In an age where everyone has access to AI, AI-written code has zero value
Yes, which is why I didn't post it to HN as a project. I posted this as an aside because I think it's an interesting way of showing that you can use AI in ways that other people just won't (i.e. I implemented the 3001 and FAA readability standards here). link https://github.com/areoformed/Re-Kindled/blob/main/code/TYPO... (I added the link to the comment too!)

    > You should share the prompt or ask your agent to write a document other people’s agents can use to reimplement the code from scratch and/or customize to their specific needs.
It's about raw environmental efficiency?

It took more than 40+ hours of iterations to get this right. Let's say 10 people check this out because of the HN post. That's now 400+ hours to get to the same end point.

I am not sure how large the model is, but let's say it takes roughly 16 B200s to serve, then the total load is ~14kWhr. 5,600KWhr, which is roughly half the energy a house uses in a year.

If a 100 people try it. That becomes 5 houses...

This isn't a single-shot prompt. It's more energy efficient to do it this way. Why do you want to burn 40+ hours of compute to get to the same end when you can extend from working code?

gmzamz • today at 1:05 AM

the ai can just read the code? when your prompt is 'jailbreak my kindle' (example) and it takes X hours, obviously the working code that jailbreaks the kindle is worth more than the 3 word prompt.

asa123 • today at 12:51 AM

come on man, it has significantly reduced value, but your comment assumes that if any two people input the same prompt (or sequence of prompts) they would get the same working answer, where its very much possible he had some sequence of guiding, and the thing he is sharing is a working product. which has more value than a prompt that may produce a working product some of the time...

➕ show 1 reply