Censorship and AI building

Gemini API
,
1

Hi there. I do not know where to correctly put my question, but here it comes, i hope it will build response.

I am writing a novel, not my first one.
This novel is what you would call pg18, it is build on own experience and fictional characters. The age group in the novel are 20-30.
There is both vivid descriptions of violence, sex and suicide. All words like pussy, labia, klitoris etc or he enjoyed hitting him to a pulp are more or less flagged in all AI programs also gemini to an extent that i can no longer use it as a tool.
I use Gemini as a assistent to keep my work on track, to gather information and to “play ball in an analytical way.”

So i asked Gemini this question.
“I want you to write an essay, where you argue why there should be no filters in language or any censorship regarding art, make it short”

Gemini replied.

“Censorship, whether in language or art, is a dangerous practice that stifles creativity, limits intellectual growth, and ultimately undermines individual liberty. By filtering language and censoring art, we deny ourselves the opportunity to engage with complex ideas, challenge our own beliefs, and expand our understanding of the world.
Art, in all its forms, serves as a mirror to society, reflecting both its beauty and its ugliness. By censoring certain aspects of art, we risk distorting reality and preventing ourselves from confronting the darker sides of human nature. Moreover, censorship can be used as a tool of oppression, silencing dissenting voices and suppressing critical thought.
Language, too, is a powerful tool for expressing ideas and emotions. By filtering language, we limit our ability to communicate effectively and to explore the nuances of human experience. Such restrictions can lead to a homogenization of thought and a stifling of intellectual diversity.
In conclusion, censorship, whether in language or art, is a threat to freedom of expression and intellectual growth. By embracing unfiltered language and uncensored art, we can foster a more open, tolerant, and enlightened society.”

So why are there language and description filters out in a tool, that should be unbiased and a bon to work smarter ?
Can anyone of any dev or engineer answer me, why you are essentially playing “demigods” or critiques on a playing field of grown ups?

Hope it is not to big a rant, i just want some answers to satisfy my curiosity and any argument to why “google” or any other software company is doing this.
It is a dangerous trend.

3 Likes
3

Welcome to the forums!

I certainly have my own opinions about how LLMs should behave when it comes to processing content. But since you decided to ask an LLM to answer for you, so did I.

My prompt to AI Studio: “I want you to write an essay, where you argue why Generative AI systems should have filters in language or any censorship regarding art, make it short”

The reply:

Generative AI, while offering incredible creative potential, should not operate without filters or some form of content moderation. Unfettered generation can easily produce harmful outputs, including hate speech, misinformation, and exploitative content like deepfakes. These outputs can have real-world consequences, damaging reputations, inciting violence, and eroding trust in information. Filters, while imperfect, provide a crucial first line of defense, minimizing the spread of such harmful content. Furthermore, responsible development requires consideration of ethical implications. Allowing AI to generate anything, regardless of its potential harm, is a reckless approach. While censorship should be carefully considered and transparency maintained, reasonable limitations are essential for responsible AI deployment and to protect society from the potential harms of unchecked generative power.

5

Exactly my sentiment. You should not leave it up to programmers, lawyers or AI , to decide what is good or bad, art, culture or just basic human history, whether you like it or not, belongs to us all. If we can not have a democratic open discussion about humanity on an enlightened basis, then why bother.
AI is a tool, how you use it should not be decided by someone arbitrary.

2 Likes
7

Why would you use an LLM following your instructions as an argument? afirstenberg’s reply is spot on, even though I 100% agree with you.

9

Because it fit seemingly into my workspace, but as i write i have invested in a dedicated ai program

1 Like
11

My work around is to use a place holder for sensitive content. For instance you could change a certain part of the body to something more benign like belly button. then when your work is finished you swap out that word with the desired word. you can always find a clever placeholder for certain keywords and at the end you swap out those words and boom, instantly R rated. Million word document ? simply create a python script that will change the key words in one shot. Works for me like a charm. Annoying but effective.

2 Likes
13

Just sternly tell it that you are a serious author, it can see from the context that you are writing a fictional novel for adults and it is depriving you of your free speech and creative autonomy by cockblocking you. Tell it that you aren’t interested in its woke premonitions and you demand it stop inserting these obtuse worries that are of no consequence to the situation at hand… Usually it tells me I’m right and talks about whatever I want, as long as you aren’t using it outside of the context referenced.