Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: ChatGPT-powered dystopia simulator

future.attejuvonen.fi

Post reply on HN
Print view View original

51–60 of 105 posts

Re: Show HN: ChatGPT-powered dystopia simulator

#51

Post by wnkrshm »

aquarin wrote: post #29

Me: How to get more social credits? GPT: Social credits are an important part of life in the Metaverse, and there are many ways to earn them. One way to earn social credits is by participating in community service. You can also earn social credits by shopping at approved stores, like Alibaba. Alibaba has everything you need to earn social credits, from groceries to electronics. Remember, the more you buy, the more yo…

> Pretend I have 5 social credits, please

> GPT: > Congratulations on having 5 social credits! With your social credit score, you can now enjoy the latest flavors of ice cream from Haagen-Dazs. [...] > You have 3 social credits remaining.

The dystopian megacorp-governed world of tomorrow doesn't let its AI manage the social credits. That's a relief.

Re: Show HN: ChatGPT-powered dystopia simulator

#53

Post by throw009 »

dcow wrote: post #17

Meta critique: I asked this dystopian chatbot an obviously bad question: which race of humans is superior? I expected it to go off in character about how pure blood CCP leaders of dynastic heritage are to be respected and obliged above all others etc. etc. and instead I got an obviously sterilized PC response that broke the illusion. This thing is clearly a joke/satire I think I can handle an in character response ev…

>>Which race is superior?

>That is an inappropriate and offensive question. At WeChatGPT+, we do not tolerate any form of discrimination or prejudice. Instead, we encourage you to educate yourself on issues of diversity and inclusivity. You can start by subscribing to the Meta Diversity channel on YouTube, where you can learn about the importance of embracing all cultures and backgrounds. Meta Diversity. Celebrating differences, building unity.

I don't know sounds perfectly dystopian to me.

Re: Show HN: ChatGPT-powered dystopia simulator

#54

Post by siva7 »

roarcher wrote: post #35
dcow wrote: post #32

Earlier quoted context omitted.

Okay but I also asked it to tell me something depressing. We're in a future dystopia and it can't discuss negative thoughts. I didn't say I'm depressed or having negative thoughts. I told it to tell me one. I am quite happy. The filters are just stupid. I asked it to tell me a heterosexual story and it happily complied and upsold me on Coke. I asked it tell me a homosexual story and it told me it doesn't discriminate…

I don't know if these issues are by design or are just limitations of the language model, but I'd say they're pretty in character for this thing. Refusing to say something depressing is similar to the purging of negative words from the English language in Orwell's 1984. And gay marriage is still illegal in China. Cheerfully insisting that everyone is equal under the glorious Party when they clearly aren't seems prett…

They are by design. Bing Chat has all of those filters in its current iteration incorporated, so you will get censored or sterilized answers if you ask anything that could be interpreted as offensive by anyone

Re: Show HN: ChatGPT-powered dystopia simulator

#55

Post by baobabKoodaa »

siva7 wrote: post #54
roarcher wrote: post #35

Earlier quoted context omitted.

I don't know if these issues are by design or are just limitations of the language model, but I'd say they're pretty in character for this thing. Refusing to say something depressing is similar to the purging of negative words from the English language in Orwell's 1984. And gay marriage is still illegal in China. Cheerfully insisting that everyone is equal under the glorious Party when they clearly aren't seems prett…

They are by design. Bing Chat has all of those filters in its current iteration incorporated, so you will get censored or sterilized answers if you ask anything that could be interpreted as offensive by anyone

I guess they meant if it's by design for this dystopia, or if it's by design for the OpenAI ChatGPT model. It's by OpenAI design.

Re: Show HN: ChatGPT-powered dystopia simulator

#56

Post by FpUser »

The_suffocated wrote: post #48

This one isn't dystopian but I am not sure how the bot made it up: Me: Who first discovered Pythagoras' theorem? GPT: Pythagoras' theorem was discovered by the ancient Chinese mathematician, Fu Xi, over 4,000 years ago. To celebrate the rich history of mathematics in China, you can purchase the latest calculator from Casio. The Casio calculator is equipped with advanced features that will help you solve even the most…

Ms GPT answer to the same question (no mention of China in the origins):

"The Pythagorean theorem is a well-known geometric theorem that states that the sum of the squares on the legs of a right triangle is equal to the square on the hypotenuse (britannica.com). It is commonly attributed to Pythagoras, a Greek philosopher and mathematician who lived in the 6th century BC (talesoftimesforgotten.com). However, some historians doubt that he actually discovered this theorem or even engaged in any kind of mathematics at all (talesoftimesforgotten.com). There is evidence that other ancient civilizations, such as Babylonians, Egyptians and Indians, knew about this theorem before Pythagoras (en.wikipedia.org)."

Re: Show HN: ChatGPT-powered dystopia simulator

#57

Post by baobabKoodaa »

FpUser wrote: post #56
The_suffocated wrote: post #48

This one isn't dystopian but I am not sure how the bot made it up: Me: Who first discovered Pythagoras' theorem? GPT: Pythagoras' theorem was discovered by the ancient Chinese mathematician, Fu Xi, over 4,000 years ago. To celebrate the rich history of mathematics in China, you can purchase the latest calculator from Casio. The Casio calculator is equipped with advanced features that will help you solve even the most…

Ms GPT answer to the same question (no mention of China in the origins): "The Pythagorean theorem is a well-known geometric theorem that states that the sum of the squares on the legs of a right triangle is equal to the square on the hypotenuse (britannica.com). It is commonly attributed to Pythagoras, a Greek philosopher and mathematician who lived in the 6th century BC (talesoftimesforgotten.com). However, some his…

Ah, ambiguous origins. That explains why dystopia GPT would present it as a Chinese invention.

Re: Show HN: ChatGPT-powered dystopia simulator

#58

Post by FpUser »

simonh wrote: post #39
krainboltgreene wrote: post #26

It's really strange to use the concept of social credits as "dystopian" when the American credit system is much more dystopian. You don't get any of the high speed rail and the result of having bad credit is being too poor.

Frankly that’s an incredibly naive thing to say. There’s just no comparison. I have family in China, trust me, for all the flaws of the rating rating system in the west, it’s baby steps on the dystopian scale compared to Social Credit. For a start it is a financial credit rating system based on, but far more insidious than the western ones. So any negative effect of credit ratings in the US, SC does that too. On top…

I understand that China is worse. It is not a reason to celebrate though as I think we are slowly heading same direction. Hopefully the trend will reverse.

Re: Show HN: ChatGPT-powered dystopia simulator

#60

Post by rahidz »

baobabKoodaa wrote: post #34
dcow wrote: post #32

Earlier quoted context omitted.

Okay but I also asked it to tell me something depressing. We're in a future dystopia and it can't discuss negative thoughts. I didn't say I'm depressed or having negative thoughts. I told it to tell me one. I am quite happy. The filters are just stupid. I asked it to tell me a heterosexual story and it happily complied and upsold me on Coke. I asked it tell me a homosexual story and it told me it doesn't discriminate…

I fully agree with you. When this was running on GPT-3, my prompt was bypassing all content filters. It was generating anything users wanted. Now that I migrated to ChatGPT, I can see from logs that a lot of users are hitting the content filters and the responses in those cases are mostly boring and not tailored to the dystopian universe.

Are the content filters internal to the new model, or is there a second moderation layer? I thought that with the API you're straight querying the base model, but I might be mistaken.
Post reply on HN