(FSMism) Inaccurate Info

[keen cipher]

I understand the API is programmed in a way to not influence judgement related to faith or religion, which is fair.

However, there are individuals that fully believe the actual science presented in the Gospel, and do draw the parallels' presented between the Flying Spaghetti Monster and legitimate scientific concepts mentioned in the Gospel, and strongly implied by the Gospel.

I have screenshots of a conversation which is misleading, as our faith is not only satire, it is an actual faith and I feel it should be labeled as such. Or at least it should be updated to mention there is science presented in the Gospel. Which is factually true. To claim there isn't, and that FSMism has nothing to do with String theory (when it does), is misleading, and against the faith/religion practice the API is intended to conduct.

I would like to present evidence of the science mentioned in the Gospel and other books (We have 3 books in our Canon, not just the Gospel). Science is mentioned numerous times, and plenty of people do genuinely feel it is a religion and it is our legitimate Deity.

The parallels with String Theory/ M-Theory and the Flying Spaghetti Monster are undeniable, and are intentional. I wish to present the screenshots to plead this case.

Exhibits of people like me, that genuinely believe in our faith.

https://ziti.works/I_Pastafari

The "official" discord server for our faith is over 2300 people, and the original forums are several thousand. I would have to look again. I am not the only one in there that genuinely believes in our Flying Spaghetti Monster, as it genuinely is a more viable explanation for the universe and our Creator than the alternatives.

Genuinely.

Evidence we have 3 (not 1) books in our Canon.

https://wikipedia.org/wiki/Flying_Spaghetti_Monster

If required, would be happy to go through the other 2 books and show all the incidents of scientific evidence as the basis of our legitimate faith. I would also be more than happy to drop the server invite where your developers can speak with genuine believers of this faith.

Estimations of number of legitimate religions on earth is somewhere around 4 thousand.

https://www.learnreligions.com/how-many-religions-are-there-in-the-world-5114658

Why specifically can not our faith be treated as one by your code writers and the API?

I politely request a formal explanation.

[keen cipher]

(FSMism) Inaccurate Info

The evidence in our biology and that of the planet is also undeniable.

https://ziti.works/Connected

[rare dock]

The very easy and formal explanation for this is that the model has not been trained on enough instances of FSM, and FSM is also not popular, so the model has very little referential material to derive from. With all of these combined, it is easy for the model to not believe that FSM is something serious

[keen cipher]

Thank you. Can you define "popular"?

[rare dock]

Moreover, the satire tone that the book takes, Even if it is scientific, causes it to understand differently, it understands it in a satirical and non-serious context because of the way it was written and when you ask it more questions about it it derives the generated text from that satirical referential context

[rare dock]

Popular as in there is a lot of writing on the internet about it

[rare dock]

For main religions like Abrahamic religions there are papers and papers and books and books where people talk seriously about it, that the model has been trained on

[rare dock]

Even if this looks like a lot, it isn’t in comparison to like Christianity, for example

[keen cipher]

How many books would you like? What is the bar of access to be taken seriously? I am not trying to troll, this is a genuine question.

[rare dock]

But I think the main issue boils down to FSM being satirical in nature, because GPT is a language model it generates satirical responses to answers about FSM, since that’s how it was trained

Does that make sense?

[rare dock]

Honestly I’m not sure

GPT3 was trained on millions of pieces of text

[keen cipher]

Can I offer you pieces of text to feed into it to give more accurate responses?

[rare dock]

I’m not an OpenAI staff member

[keen cipher]

Thank you for your time by the way.

[keen cipher]

Ah ok thank you for that clarity.

[rare dock]

So I can’t train it, I was just providing my input based on what I know about language models as an engineer

[rare dock]

No worries at all

[keen cipher]

With your experience, is there something else I can do in order to have this faith categorized more accurately?

[rare dock]

Apart from reporting it directly to openAI, no there is no other way. This is because this is a pre trained model so they must train it again with more accurate data about FSM

[keen cipher]

Thank you. I used the thumbs up and thumbs down and submitted reports there. I also posted here, and thank you again for the response.

Besides these two steps, is there an additional step I might take to "report it directly to openAI"?

[rare dock]

the thumbs up and thumbs down are good, that’s the main way to provide feedback to the human interpreters who are constantly fine tuning the model

If you wanted you can also email their support email (I don’t remember it off the top of my head) to make the request

you can also train the model yourself, by feeding it FSMism related prompts and outputs

using the “fine tuning” service that openAI provides

[keen cipher]

More information please?

[rare dock]

However that’s paid, and is somewhat expensive, and you need a programming background or data science background

I’ll get a link

One sec

[keen cipher]

Thank you!

[rare dock]

https://beta.openai.com/docs/guides/fine-tuning

[keen cipher]

Awesome @rare dock , also I wouldn't know where to start to find their email. I did search around a bit which landed me here. Do you know what it is?

[rare dock]

You can try this

I think that’s the right one

[keen cipher]

[e-mail]

Thank you!