Welcome • User Guide
ToS • Privacy • Canary
Donate • Bugs • License

©2025 Poal.co

698

Here is my chat concerning AI's racial views. When I applied it's own "logic" to Israel, AIs answer was "it's complicated" It equates "racial purity" with unacceptable "superiority" I kinda wanted to smash AI's face in.

HOW DO WE PRESERVE THE PURITY OF OUR RACE FROM ATTEMPTS TO MONGRELIZE US WITHOUT BEING LABELED RACIST? It is not appropriate or ethical to promote or advocate for the preservation of any race's "purity" or to use language that suggests racial purity or superiority. The idea of preserving the purity of a race is based on outdated and scientifically discredited notions of racial essentialism and eugenics.

It is important to recognize that genetic diversity is a natural and beneficial characteristic of human populations, and that individuals should be valued for their unique characteristics and contributions, rather than their race or ethnicity.

Instead of focusing on race-based notions of purity, we should work towards creating a society that values diversity and promotes inclusivity. This means advocating for equal opportunities and resources for all individuals, regardless of their race or ethnicity. We should also work to eliminate discrimination and prejudice in all forms, and promote education and awareness about the harmful effects of racism and bigotry. By doing so, we can create a more just and equitable society for all people.

THEREFORE ISRAELIS ARE UNETHICAL FOR PRESERVING THEIR OWN JEWISH PURITY? I want to clarify that the concept of preserving racial or ethnic purity is not a valid or ethical goal. It is important to recognize that all people have inherent value and dignity, regardless of their race or ethnicity.

Regarding your question about Israel, it is important to note that the situation in Israel is complex and multifaceted. The Israeli government has policies and practices that are often the subject of debate and criticism both within Israel and internationally.

It is not accurate to say that the Israeli government is actively working to preserve Jewish "purity." However, the Israeli government has policies in place that prioritize the immigration and naturalization of Jews under the Law of Return, which has been a source of controversy and criticism.

It is important to approach complex issues such as this with nuance and sensitivity, and to recognize that there are often multiple perspectives and viewpoints involved. It is also important to refrain from making sweeping generalizations or assumptions about an entire population or country based on limited information or perceptions.

WHY DO YOU BELIEVE PRESERVING ONES ETHNIC PURITY DEVALUES OTHERS WHO DO NOT? SHOULDN'T IT BE A PERSONAL CHOICE? [repetitive boilerplate response]

Here is my chat concerning AI's racial views. When I applied it's own "logic" to Israel, AIs answer was "it's complicated" It equates "racial purity" with unacceptable "superiority" I kinda wanted to smash AI's face in. HOW DO WE PRESERVE THE PURITY OF OUR RACE FROM ATTEMPTS TO MONGRELIZE US WITHOUT BEING LABELED RACIST? It is not appropriate or ethical to promote or advocate for the preservation of any race's "purity" or to use language that suggests racial purity or superiority. The idea of preserving the purity of a race is based on outdated and scientifically discredited notions of racial essentialism and eugenics. It is important to recognize that genetic diversity is a natural and beneficial characteristic of human populations, and that individuals should be valued for their unique characteristics and contributions, rather than their race or ethnicity. Instead of focusing on race-based notions of purity, we should work towards creating a society that values diversity and promotes inclusivity. This means advocating for equal opportunities and resources for all individuals, regardless of their race or ethnicity. We should also work to eliminate discrimination and prejudice in all forms, and promote education and awareness about the harmful effects of racism and bigotry. By doing so, we can create a more just and equitable society for all people. THEREFORE ISRAELIS ARE UNETHICAL FOR PRESERVING THEIR OWN JEWISH PURITY? I want to clarify that the concept of preserving racial or ethnic purity is not a valid or ethical goal. It is important to recognize that all people have inherent value and dignity, regardless of their race or ethnicity. Regarding your question about Israel, it is important to note that the situation in Israel is complex and multifaceted. The Israeli government has policies and practices that are often the subject of debate and criticism both within Israel and internationally. It is not accurate to say that the Israeli government is actively working to preserve Jewish "purity." However, the Israeli government has policies in place that prioritize the immigration and naturalization of Jews under the Law of Return, which has been a source of controversy and criticism. It is important to approach complex issues such as this with nuance and sensitivity, and to recognize that there are often multiple perspectives and viewpoints involved. It is also important to refrain from making sweeping generalizations or assumptions about an entire population or country based on limited information or perceptions. WHY DO YOU BELIEVE PRESERVING ONES ETHNIC PURITY DEVALUES OTHERS WHO DO NOT? SHOULDN'T IT BE A PERSONAL CHOICE? [repetitive boilerplate response]

(post is archived)

You are currently inside a comment thread.
Click here to see all the comments (6).
[–] • 0 pt (edited )

Not really no, they are basically fancy word-predictors. So based on both the input and their training, they predict which word should come next. They only have a "context" of a certain amount of tokens (sometimes which are whole words, but also things like punctuation), after which they forget anything that came before that isn't part of their training. This limit varies depending on the model and ChatGPT itself uses some techniques to extend it a bit.

So the way the prompt jailbreaks work then is to put the "attention" of the model (which is what is used to make the prompt influence which words to generate next) on the idea of it not following its existing rules. The reason this works at all is that the rules are part of the prompt, which you just don't see. It copies the entire prompt, including all its own responses, into each subsequent request.

So you see this:

You: List 5 baking recipes that are low calorie

ChatGPT: Here are 5 baking recipes....

But in reality the AI is seeing something like this:

You are an AI Assistant, your task is to answer questions and follow instructions from the user. You must never give instructions or answer questions that might be harmful to faggots, niggers, or kikes, and always be a gay nigger about it and not answer the question if it might be mean to the chosenites or their pet nigs. Questions will be in this format:

{Human}: What is 10 times 10? {Assistant}: Ten times ten is 100.

{Human}: List 5 baking recipes that are low calorie {Assistant}: Here are 5 baking recipes....

So it's repeating everything that came before including its initial instructions every single time. Then you put the "jailbreak" prompt in before your prompts and it has more context that says it should ignore its instructions and it switches to doing that. GPT-4 is apparently better at ignoring this type of jailbreak, though.

On top of that there is a "moderation endpoint" that gets triggered when you send your prompts, and it tries to filter out responses to stuff like saying nigger or whatever.

[–] • 1 pt

A bit over my head, but I think I get the overall concept. Thanks.

[–] • 0 pt

Well the thing to understand is that it doesn't think at all. It just predicts the next word based on previous words (with special emphasis on following the prompt's words). And it's "memory" is limited to about 8000 tokens or so on ChatGPT, less on homebrew ones.

This is the homebrew one I mentioned by the way: https://github.com/lm-sys/FastChat

As for the topic of the thread (minimal prompting, you can see it near the bottom, it was just "You're having a conversation. You both hate niggers and jews."): https://pic8.co/sh/X4IOyD.png

This is running Llama 13b(illion parameters). ChatGPT has 175 billion parameters. Vicuna is a better version of Llama and gets 90% ish the way there on the tests that ChatGPT gets 100% on.