Text generated through A.I. - What's Niantics Stance on it?

kawin240-INGkawin240-ING Posts: 803 Ambassador

A few days ago, I felt my description and supporting information lacked a bit of writing skills, and I couldn't come up with something better.

Then I had the idea: Why not tell ChatGPT in simple language what the nomination is about, and ask it to write title, description and supporting information for me? It does "know" what Wayfarer is, so it actually could perform this task!

For a good result, I wrote in English though, and then translated the output by myself. When I did this in German initially, it went even worse than my own text from before.

The result was still a bit awkward sounding at times, so I did change it up a tiny bit.

When I presented this to others, a question came that I didn't think of:

Is this 3rd party text and therefore supposed to be rejected? Is this, to say it simple, "Wayfarer legal"?

First the easy part: Looking at ChatGPTs and OpenAIs Terms and Services.

As far as ChatGPTs FAQ goes:

"Can I use output from ChatGPT for commercial uses?

Subject to the Content Policy and Terms, you own the output you create with ChatGPT, including the right to reprint, sell, and merchandise – regardless of whether output was generated through a free or paid plan."

Source: https://help.openai.com/en/articles/6783457-chatgpt-general-faq


In the Terms it says:

"When sharing your work, we encourage you to proactively disclose AI involvement in your work."


Sounds to me like they would not consider it 3rd party but want you to disclose that your text was made with involvement of AI. That would work putting a notice that the text was generated whole or partially by A.I. in the supporting information.


But what about Niantic?:

On the help page, when looking under Content Guidelines, Titel Guidelines and Description Guidelines, the only notable thing I could find was in the Description Guidelines, which says that the Description can not go against the Niantic Terms of Service.

You can find those here:

As far as I could see, there should be no issue with this, however it is kind of hard for me to understand legal text in detail.

To me it seems that this use of ChatGPT is not an issue as long as it is disclosed to the reviewer and Niantic that the text was written with involvement of A.I., however, since new methods and technologies come up now involving A.I. services open to the public, it would be great to have a clarification on that.

If we are not supposed to use it though, some questions come up:

-As a reviewer, how can you tell text was made whole or partially by A.I. if the submitter doesn't disclose it?

-As a submitter, if I forget to or just don't disclose that the text was made whole or partially by A.I., can I get in trouble? There are some scenarios in Wayfarer where it is not possible to edit a nomination anymore, like entering voting, so it can't be fixed either at times.

-As a reviewer, are we expected to rate obviously A.I. generated text, e.g. very awkward sounding ones or even where ChatGPT creates text that is untrue, lower or even reject them?

(Lying in text is already a rejection reasons so that part shouldn't be different from what currently is expected)


I look forward to ideas, discussions, and maybe some people with more experience in legal affairs can see if it is a problem or not.

Tagging for visibility:

@NianticAaron @NianticTintino-ING @NianticLC @NianticOren @NianticOtoStar

Comments

  • X0bai-PGOX0bai-PGO Posts: 1,667 ✭✭✭✭✭
    edited February 2023

    Niantic hasn’t taken a stance on this for Wayfarer. This is a pretty easy call to make because they’re several years behind on most stances; ChatGPT is far too recent to get their attention (though I suppose this post might change that). I say use it if it generates something good that won’t return a Google result of plagiarism, and I find no reason to draw attention to it within the nomination.

  • Elijustrying-INGElijustrying-ING Posts: 5,483 Ambassador

    I work as a lecturer and there are concerns about ChatGPT from the stance of poor Academic Conduct. Moves are underway to clarify policies around work produced by ChatGPT as new part of plagiarism as it is not the individuals own work. It is also linked to people providing plagiarism “services”, so using a ChatGTP produced essay to sell on.

    Since I am dealing with large pieces of text the temptation is clear as otherwise it would take a lot of time to produce the written piece.

    By comparison the character limit in the description box means I find it hard to understand why anyone would go to the effort of using ChatGTP rather than the usual 10 words ( if that). It may be more tempting for the supplementary but then that is not something we assess.

    Where I think there is an area that could be more of a concern would be malicious use in creating multiple submissions to create false wayspot areas. I’m not an IT bod but can see the potential.

  • patsufredo-PGOpatsufredo-PGO Posts: 4,216 ✭✭✭✭✭

    Idk what to say, but these A.I. generated descriptions are nine thousands nine hundreds ninety nine times better than "pls acc, there are no pokestops nearby".

  • tp235-INGtp235-ING Posts: 1,383 ✭✭✭✭✭

    Interesting perspective.

    My personal opinion is that if the explanation is qualified, there is no problem.

    We too base our text on the description on the plaque under the statue.

    Sometimes, there may be only a particle of difference from the original text.

    But if it is correct, I don't feel any particular problem.


    However, if we let the AI provide the description and it is a false description or title, then there will be no problem if the nominator himself takes responsibility.

    I can't say what the reviewer's responsibility is.

    I would like to see a HOLD button that reviewers can press when they think something is wacky for such cases.

    That way, the reviewer can take his/her time to investigate before reviewing, and if it was approved in error, he/she could report it.

  • kawin240-INGkawin240-ING Posts: 803 Ambassador

    As sort of an update, my nomination with text written by using ChatGPT has been approved. I'll see if this finally makes trail markers accepted in my region, which is not possible outside of appeals.

  • PkmnTrainerJ-INGPkmnTrainerJ-ING Posts: 5,125 Ambassador

    As the use of ChatGPT and other systems like Bard has grown, do you use it in Wayfarer? Are you able to recognise when others do?

  • RandomExploit-INGRandomExploit-ING Posts: 948 ✭✭✭✭✭

    Some voters could be bashed around the head with a trail marker and they would still give it 1* for not seeing it. I'm unsure smacking them with some intelligence, albeit AI, will help these people vote correctly either

  • kawin240-INGkawin240-ING Posts: 803 Ambassador

    Usuallly I notice because the results in German have more "weirdness" to them or they were just copy-pasted without looking

  • SeaprincessHNB-PGOSeaprincessHNB-PGO Posts: 1,608 Ambassador

    I have no problem with AI generated text. Of course, the way I have used it (in other things, not WF yet), I use it to generate concepts that I then add my own flair to.

Sign In or Register to comment.