Re: GPT tool for Checking Challenge Cache Compliance with Guidelines

Apologies in advance if this is not the correct place to share this. The tool is not a script but a simple tool designed to check whether a challenge cache idea complies with the current guidelines. I wasn’t sure where this belongs, so I posted it here.

To reduce the number of non-compliant requests, it might be beneficial to add something like this directly to the “Checker Requests (Read Me First)” forum section.

This is just a prototype where a GPT model compares the challenge idea against the guidelines. While it’s not entirely foolproof, it can be further improved.

The only requirement is to be logged in to OpenAI with a registered account. It should work for everyone and is free to use (you only need an OpenAI account).

https://chatgpt.com/g/g-678c2e0abbac81919722e49750b83e7b-gc-challenge-validator

"Is a streak challenge for 500 days allowed?"

ChatGPT said:

A streak challenge requiring 500 consecutive days of geocaching is allowed under the Geocaching Challenge Cache Guidelines, but it must comply with certain conditions to ensure it aligns with the rules set as of June 23, 2021
...

Literally the first question I asked, and it's just wrong. The guidelines aren't that hard to follow... The problem is that people just aren't reading them at all, and this tool might give someone the false sense that their challenge idea is allowed, when it definitely isn't.

Great example of why it's stupid to use ML for things like this.

Okay, that makes sense too. I’ll consider this matter closed.

The guidelines might seem straightforward if you’re familiar with them—you remember definitions and terms like “streak” without any issues, not to mention the English language itself. However, I come across many people who have an idea but, despite having access to the guidelines, only translate part of them with their clumsy translation and may not fully understand them. So, it’s probably not as straightforward as it seems. Based on what I sometimes see in the forum’s script requests, quite a few push the boundaries or even go beyond what’s allowed.

Regarding your attempt, even with precise input, GPT often requires fine-tuning, but that’s the case with almost everything in the IT world. This was just a suggestion; I’m not forcing anyone to use it. :)

I'm in IT (although I'm snobbish and would probably not call my job an "IT job")... and honestly, I don't think every problem is best solved by technology.

As I said, I think it's bad to have it available, because it will give inaccurate answers (at least sometimes), and I'm sure some users will use its output as a justification that their challenge is allowed, when it isn't. The guidelines have translations into many different languages, and if you can't understand them (but more likely can't be bothered to read them at all), you probably won't be using the ChatGPT bot either. We've got many active people on here who are very knowledgeable on the guidelines, including the "unwritten rules" which take into account regional variations... and will politely let you know if you (accidentally :)) request a challenge that isn't allowed.