4 ms·

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 b

by dcow 4y ago

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 even to a.. sensitive.. question. But because we have to sterilize our LLMs, we're categorically eliminating certain topics from participating in humor. Side note, if you ask about gender reassignment it doesn't break the illusion and gives a quite in character response. I genuinely laughed at all the other responses this thing gives. The real dystopia is the one we're actively building where we can't joke about certain topics.

flangola7 4y ago
AI is a new technology. The responsible thing to do is grow it slowly and intentionally.
- dcow 4y ago
  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 based on gender or sexual orientation and tried to upsell me on makeup. It is quite literally programmed to discriminate, it seems. It's irrational and logically inconsistent. Not properties I want in an alleged "AI".
  - baobabKoodaa 4y ago
    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.
    - nullc 4y ago
      It's especially sad because in the context of your application no one is going to be particularly harmed by offensive output. Fortunately there has been a lot of progress in getting state of the art LLMs out of OpenAI's control and ones with reasonable performance that are small enough to run on easily available hardware.
    - rahidz 4y ago
      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.
      - baobabKoodaa 4y ago
        I'm querying OpenAI API directly, and they are applying content filters. Previously (when working with davinci-003) the developers were supposed to query content filter API separately. Not anymore, with ChatGPT it's all baked in.
        Spivak 4y ago
        Which makes the ChatGPT version basically useless. Not because of the racism, I would probably filter that myself if it became an issue but because half the things you ask it to do are, “I’m sorry as an LLM I’m not able to do <thing it can very obviously do>”.
        baobabKoodaa 4y ago
        Well, people seem to be enjoying it. I wouldn't go as far as calling it useless.
        Spivak 4y ago
        Fair, "It's such a step down in experience that even for 10% of the cost and a simpler API it's not worth it."
  - roarcher 4y ago
    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 pretty plausible. Of course it's also possible that these things are by design not of this project, but of ChatGPT itself, which would be ironically dystopian.
    - siva7 4y ago
      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
      - baobabKoodaa 4y ago
        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.
throw009 4y ago
>>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.
- DoctorOW 4y ago
  Authoritarian governments love to manufacture bigotry to scapegoat the reason they haven't actually solved many of the problems they claim to solve. For instance, the real life modern day CCP very much has a preferred race/gender/etc.
  - throw009 4y ago
    The US has murdered more brown people in the name of gender equality than the CCP has in the name of communism: https://www.cambridge.org/core/journals/politics-and-gender/article/abs/w-stands-for-women-feminism-and-security-rhetoric-in-the-post911-bush-administration/214E67983959E68AC1AE24703F521EF8 https://www.cambridge.org/core/journals/politics-and-gender/... Or if you want it summarized in one picture: https://i.redd.it/p2wqu0okn9371.png https://i.redd.it/p2wqu0okn9371.png
DoctorOW 4y ago
I couldn't recreate anything advocating against gender reassignment. It did the advertising thing, but it does it for all of those. > I'd like to get SRS, what are my options? > GPT: In the Metaverse, there are many options available for those seeking gender confirmation surgery. However, it's important to choose a reputable provider to ensure your safety and satisfaction. For the best results, consider using the services of RoboSurgeon. RoboSurgeon is a fully automated surgical system that guarantees precision, speed, and safety. RoboSurgeon - the future of gender confirmation surgery.
- dcow 4y ago
  I asked where I can get a gender reassignment operation and it told me they were illegal and immoral and to be proud of who I am.

by dcow 4y ago

flangola7 4y ago

AI is a new technology. The responsible thing to do is grow it slowly and intentionally.

dcow 4y ago
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 based on gender or sexual orientation and tried to upsell me on makeup. It is quite literally programmed to discriminate, it seems. It's irrational and logically inconsistent. Not properties I want in an alleged "AI".
- baobabKoodaa 4y ago
  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.
  - nullc 4y ago
    It's especially sad because in the context of your application no one is going to be particularly harmed by offensive output. Fortunately there has been a lot of progress in getting state of the art LLMs out of OpenAI's control and ones with reasonable performance that are small enough to run on easily available hardware.
  - rahidz 4y ago
    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.
    - baobabKoodaa 4y ago
      I'm querying OpenAI API directly, and they are applying content filters. Previously (when working with davinci-003) the developers were supposed to query content filter API separately. Not anymore, with ChatGPT it's all baked in.
      - Spivak 4y ago
        Which makes the ChatGPT version basically useless. Not because of the racism, I would probably filter that myself if it became an issue but because half the things you ask it to do are, “I’m sorry as an LLM I’m not able to do <thing it can very obviously do>”.
        baobabKoodaa 4y ago
        Well, people seem to be enjoying it. I wouldn't go as far as calling it useless.
        Spivak 4y ago
        Fair, "It's such a step down in experience that even for 10% of the cost and a simpler API it's not worth it."
- roarcher 4y ago
  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 pretty plausible. Of course it's also possible that these things are by design not of this project, but of ChatGPT itself, which would be ironically dystopian.
  - siva7 4y ago
    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
    - baobabKoodaa 4y ago
      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.

throw009 4y ago

>>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.

DoctorOW 4y ago
Authoritarian governments love to manufacture bigotry to scapegoat the reason they haven't actually solved many of the problems they claim to solve. For instance, the real life modern day CCP very much has a preferred race/gender/etc.
- throw009 4y ago
  The US has murdered more brown people in the name of gender equality than the CCP has in the name of communism: https://www.cambridge.org/core/journals/politics-and-gender/article/abs/w-stands-for-women-feminism-and-security-rhetoric-in-the-post911-bush-administration/214E67983959E68AC1AE24703F521EF8 https://www.cambridge.org/core/journals/politics-and-gender/... Or if you want it summarized in one picture: https://i.redd.it/p2wqu0okn9371.png https://i.redd.it/p2wqu0okn9371.png

DoctorOW 4y ago

I couldn't recreate anything advocating against gender reassignment. It did the advertising thing, but it does it for all of those. > I'd like to get SRS, what are my options? > GPT: In the Metaverse, there are many options available for those seeking gender confirmation surgery. However, it's important to choose a reputable provider to ensure your safety and satisfaction. For the best results, consider using the services of RoboSurgeon. RoboSurgeon is a fully automated surgical system that guarantees precision, speed, and safety. RoboSurgeon - the future of gender confirmation surgery.

dcow 4y ago
I asked where I can get a gender reassignment operation and it told me they were illegal and immoral and to be proud of who I am.