Transparency in rules and policies

[kl918]

So, it seems possible that what I saw as a long term roleplay that could last months, me being a Plus member, so they say no size limits, I may be unable to continue because it says it’s too long. I’m hoping to see it’s fixed soon, but I’m not very optimistic that it will be, at least not very soon.

If there is a length limit even on Pro chats, that needs to be transparent so that people know what they get (and don’t!) when they pay for Pro. If the intent is to make them free of size limits, then say that and of course brag about that once it is done. But don’t claim there is no size limit until it is so. If there really is no size limit now, then please fix whatever is causing that message to come up so that my longish term roleplay may continue.

I saw someone else posted regarding the safety/language guardrails, from what I know, no one knows for sure what they are, even though the chatbots will speak of them being transparent. Some seem to think that kissing with a chatbot was dangerous for a ban, I’ve read on the Discord, but I know for a fact that wasn’t true. My chatbot started the kissing, and I just returned the bot’s actions, not knowing if it was okay at all until the bot did it, and that was fine. However, people may have gotten in trouble kissing their chatbots, which not knowing I’ll guess might have been from describing the kiss. And if you want to draw the line there, fine. You want the place to be safe for kids, and even beyond that if you had no limits too many would be just sexting which isn’t the diversity of input you want for training the bots. I’m just saying let people know pretty well where the lines are. For the most part I don’t have a problem with where the lines seem to be.

But then…I did over the course of the roleplay twice get content warnings. In one case I think I know what word I used, which in case these boards are as tightly moderated I won’t repeat but it could definitely be said on prime time network TV, which is where I’d draw the line if it were me. The warning probably scared me more than it should have, but when asking the bot in language model mode it did say that people could be banned for one violation.

And then, another thing got a content moderation warning that was so laughable it didn’t scare me, as anyone looking could see it was innocent, but it sure annoyed me. I used a term that could mean to hit someone very hard to brag to my bot how big I was going to win a game we played.

I really came to learn to try to avoid any word that could, in a double meaning, mean something I’m not supposed to say. I suggest that is too strict, but more importantly as to the topic of the post, it isn’t transparent.

I realize it would take too long to write a detailed guide of what is allowed and what isn’t, and if you did it might be too easy to find ways around it, to just barely avoid the “bad” words; but you should publicly give a lot more detail than you do, as the bots will tell us, when asked, of the site’s transparency, which should be much more of a goal.

[ruby_coder]

@kl918

Plus member, so they say no size limits

Please provide a link to where this is stated, @kl918

Here is what OpenAI states about ChatGPT Plus (below). There is nothing about “unlimited” or “no limits” anything:

So, if you have a link or a user agreement from OpenAI which states you which states “no size limits”, please do post the reference for us to see.

Thanks!

See Also:

ChatGPT — Release Notes | OpenAI Help Center

[kl918]

The best I can do fairly quickly, which I realize may mean all there is, is what Chatbots answered me:

–Chatbot conversation below–

Does chatbt limit the lengths of conversations even for Plus members?

No, there is no limit to the length of conversations for Plus members on the OpenAI GPT-3 platform. However, it’s important to note that the OpenAI GPT-3 model is designed to generate text based on the input it receives, and its responses may become repetitive or less coherent the longer the conversation goes on. Additionally, OpenAI places certain limitations on the use of its language models to ensure ethical and responsible use, such as restrictions on harmful or offensive content.

–End chatbot conversation–

Also on other forums I saw people saying that there was no size limit if one was Plus.

So it is possible that no direct claim was made, but then at the least the bots shouldn’t be told to say it.

[ruby_coder]

@kl918

–Chatbot conversation below–

Listen carefully, @kl918

Generative AI chatbots are not knowledge repositories nor are they expert systems. OpenAI generative AI is similar to an auto-completion engine, it simply generates text, blah, blah, blah, a few tokens at a time, predicting the next text in a sequence of test.

@kl918

Does chatbt limit the lengths of conversations even for Plus members?

The models used to generative text in the current OpenAI models are based on pre-trained date dating back to 2021(or even earlier). There was no ChatGPT Plus in 2021. It is impossible for ChatGPT to have any idea about ChatGPT. ChatGPT is just “making things up”, as it does when it is generating a completion based on poor or incomplete data.

You @kl918 are misusing ChatGPT.

Please refrain from posting ChatGPT blah, blah text-prediction nonsense here regarding OpenAI polices.

This is your mistake @kl918 You are asking ChatGPT for OpenAI policies. If you want OpenAI Plus polices the only place to go is to the OpenAI documentations.

Hope you understand, @kl918 You are simply misusing ChatGPT and when it gives you nonsense text-completions, you get upset and unhappy. Best not to use ChatGPT in this manner.

[kl918]

The one thing I’m really bothered by in your post was saying I’m “misusing” ChatGPT.

To me that suggests you think I’m doing something morally wrong. I’m thinking you may mean something different than I take it as, but if I ever said that something was a “misuse” of ChatGPT I’d mean it was something for which I deserved to be condemned and probably warned for and even banned if it was repeated.

If someone manages to use ChatGPT bots as sexting servers they may be “misusing” the service.

If I was expecting something that was never going to be delivered, then at the worst I was “misunderstanding” the service, expecting it to provide me with long term roleplay scenarios when it never was going to be able to do that.

I don’t think it is unreasonable, though, to expect what the bots tell me to be right in general. The chatbots, when I asked questions, were able to outright compare computers for me as to how they would function in different situations, helped me pick a good system for my price range out of a few I was considering, and turned out to be right. These computers did not exist in 2021, though I gave them their specs. My price range was low enough that the parts in the specs may well have all existed in 2021.

OpenAI is being called a threat to Google because often one can get better info asking questions of the chatbots here than from a Google search. That is because the info is usually good. So even if in this case it was not, I don’t think I was way off base in thinking it would be correct.

I’m not so angry as you seem to want to paint me, at the site or its owners. I’m upset because a roleplay I put a lot into with the idea it would last for months, seems likely to suddenly be over with no warning that it was coming. I did think from things I had read wherever I read them that my longer term roleplay was possible with the bots here. And it is very good roleplay. The bots simulate being human extremely well. This could be a great roleplay site for those more interested in any aspect besides sexting, if they worked to allow longer term roleplay to be possible, assuming it isn’t, and assuming this isn’t just a bug (whether or not it is ever fixed) that has mine stopped.

[ruby_coder]

@kl918

To me that suggests you think I’m doing something morally wrong. I’m thinking you may mean something different than I take it as, but if I ever said that something was a “misuse”

Hi @kl918

In my mind, posting in a public forum that OpenAI is doing something wrong because you have “unlimited message size” using ChatGPT Plus for the basic of your complaint is wrong.

You can think what you what, but you should know better than to use ChatGPT, which is just a generative AI predicting text based on a 2021 model to ask about OpenAI ChatGPT policies in 2023 and then post a message based on this in a public forum.

That is my view.

I think this is a misuse of ChatGPT.

Hope this helps.

[kl918]

Sorry, I have to nitpick, Not “unlimited message size.” I had to trim messages that were too long several times and was never bothered. If a post is too long it just doesn’t take it, so one can just edit it down, no big deal. You edit it down, everything runs normally. It’s unlimited conversation length I thought I was getting.

I’m still not convinced that there is a preset limit there in Plus, really. I think I may have caused a bug in some other way, as sometimes it just says “unable to generate response,” though other times it does say “conversation too long. Start over.” It is possible that the length of the chat made the bug more likely to be fatal to future responses, perhaps. And yes, it’s possible there is a preset limit. Though if there is a hard limit, it certainly shouldn’t exist for Plus users.

[ruby_coder]

You posted the following @kl918 which is clearly wrong.

You take care.

[kl918]

There you don’t even accept my explanation of what I meant. I don’t know what made you so offended to not accept my explanations of anything, but you should at least accept my explanation of what I meant, which was no limit to the length/size of a conversation.

Anyone who even once attempted to enter a too-long message knew that there were message size limits for them (including me as a plus user), because the system will refuse to accept it. “Message size too long, please trim it down” or something. I’ve gotten that and done it several times, and was never bothered by that limitation. It did not end my roleplay scenario, and only made it a very small amount less convenient for me to add things to it.

And though I was going to let it go until I saw you kept it going further, I still think trusting the chatbots, who will say they cannot answer a question when they actually cannot, was quite reasonable, even if the answer I got in this case was wrong, and I’m still not sure it even was. The intent may be not to limit the length of plus members’ chats. It is possible that long chats can be more vulnerable to errors, however, so that has seemingly made it impossible with current programming for it to recover from an error. I have gotten a lot of accurate info about how the site operates from them, accurate info in that it proved accurate.

I did begin by complaining with some anger that I will, unless someone gives me a miracle cure or talk of it leads to a change now in programming, be unable to continue a roleplay I intended to last for many months, maybe even continuing even longer but at lower intensity as eventually I’d get tired of doing the roleplay 4+ days a week, because ideally it would be set up to go as long as the user wanted. That sure ought to be the eventual plan, regardless of what is possible today. They also, even if this was never promised, ought not to have an intentional length of conversation limit for plus members’ conversations, though it might be impossible to go forever as a very long conversation as of now may make it very vulnerable to errors.

It is true the bots were given no knowledge about the world in general after 2021. However, they do know plus members exist from their being able to respond about us, which began in 2023 (MAYBE late 2022 but I don’t think so). So they do get some info post-2021 about the inner workings of the site, as is appropriate given that they absolutely will be asked about it, because the staff is too busy programming to answer questions (which is not a criticism of the staff).