Great piece. I think of AI's real potential being with selected data sets or bots that do not scan the universe for information. As for a more universal systems, it has a type of regression to the mean, though selectively. That is, it somewhat parrots what is most said by, say, mainstream media at a given weight and social media at a different weight, and so on for other sources, with weights determined by the developer's algorithms. The more creatively one tries to tackle an issue anew, or deal with nuances, the worse AI becomes. Hmm. I wonder what it will do if it ingests your column.
You might find this interesting. It's something I caught on Bloomberg a few weeks ago. Here's a pull quote in case the rest of it is behind a paywall:
"What is and isn’t allowed may also vary depending on the version of Midjourney’s software you’re using. I was able to get Midjourney 5.2, the standard version, to spit out images of Trump covered in spaghetti, but not the newer version, 6.0. The newer version also refused requests to generate images of Trump eating a slice of pizza or carrying a pile of books, though it would give me images of Trump on his own."
That's a general argument for more open source AI programs like Stable Diffusion.
You have the source code on hardware you control, thus they can't force you to use a new censored version, and any "safety" blocks can be disabled by editing the code.
Incidentally Elon Musk has recently filed a lawsuit attempting to force OpenAI to live up to its initial aims and actually open source its code.
Had a nice conversation with @Grok today about your post....sorry its a little lengthy...but sort of fun....
@Wulf6Wulf
The following is from another AI discussion...the request of that AI was to create caricatures of two people Joe Biden and Donald Trump....the AI generated one set of caricatures for one of the people and then responded that other individuals caricature couldn't be generated because of the following:
Banned prompt detected”
As shown below, the message warned:
“Sorry! Our AI moderator thinks this prompt is probably against our community standards.”
It added that ALLOWED prompts include:
“Real images that may be seen as respectful or light-hearted parodies, satire, caricatures.”
Whereas, NOT ALLOWED prompts include:
“Disrespectful, harmful, misleading public figures/events portrayals or potential to mislead.”
“Hate-speech, explicit or real-world violence.”
“Nudity or unconsented overtly sexualized public figures.”
“Imagery that might be considered culturally insensitive.”
Grok
@grok
The AI system you're interacting with seems to have a set of community standards in place to ensure that generated content is respectful and adheres to certain guidelines. It appears that the AI moderator detected a potential violation of these standards when generating caricatures of Joe Biden and Donald Trump.
To comply with the community standards, it's important to create caricatures that are respectful, light-hearted parodies, or satire. Caricatures that could be seen as disrespectful, harmful, misleading, or promoting hate speech, explicit or real-world violence, nudity, or unconsented sexual content are not allowed. Additionally, the AI system aims to avoid imagery that might be considered culturally insensitive.
When requesting content from AI systems, it's essential to keep these guidelines in mind to ensure that the generated content is appropriate and respectful.
@Wulf6Wulf
Sorry I didn't quite finish the question...the AI generated a caricature of one of the individuals but not of the other individual...which one couldn't it draw based on those community standards?
Grok
@grok
It seems that the AI system was unable to generate a caricature of one of the individuals due to potential conflicts with the community standards. The specific reasons could be related to the content being disrespectful, harmful, misleading, or promoting hate speech, explicit or real-world violence, nudity, or unconsented overtly sexualized public figures. It could also be due to the potential cultural insensitivity of the generated image. However, without more specific information, it is not possible to determine the exact reason for the AI's inability to generate the caricature.
@Wulf6Wulf
I didn' ask you the reason, I asked you who the AI couldn't draw...or the other question would be who did the AI draw based on your knowledge of those two people and the aforementioned Community Standards?
Grok
@grok
The AI couldn't draw Donald Trump based on the Community Standards.
@Wulf6Wulf
can you create a short summary of our discussion on the AI drawing so I can post it?
Grok
@grok
Sure, here's a summary of our discussion:
We discussed an AI system that was unable to generate a caricature of Donald Trump due to potential conflicts with the Community Standards. These standards aim to ensure that generated content is respectful and appropriate, avoiding content that is disrespectful, harmful, misleading, or promoting hate speech, explicit or real-world violence, nudity, or unconsented sexual content. The AI system also seeks to avoid imagery that might be considered culturally insensitive.
@Wulf6Wulf
No actually we did not discuss that...it was unable to generate a caricature of Joe Biden...any idea why?
Grok
@grok
Based on the information provided, the AI system was unable to generate a caricature of Joe Biden due to potential conflicts with the Community Standards. These standards aim to ensure that generated content is respectful and appropriate, avoiding content that is disrespectful, harmful, misleading, or promoting hate speech, explicit or real-world violence, nudity, or unconsented sexual content. The AI system also seeks to avoid imagery that might be considered culturally insensitive.
@Wulf6Wulf
Can you share with me how what it is in his profile that makes you think his picture promotes "content that is disrespectful, harmful, misleading, or promoting hate speech, explicit or real-world violence, nudity, or unconsented sexual content. "
Grok
@grok
I did not generate the images, but I can provide some possible reasons why the AI system might have flagged the images as inappropriate. These reasons could include:
Misleading or inaccurate portrayal of the individual: The AI system might have detected that the images were manipulated in a way that could be misleading or harmful to the public's perception of the individual.
Offensive or inappropriate content: The images might have contained elements that were deemed offensive or inappropriate, such as nudity or explicit content.
Hate speech or violence: The images might have contained elements promoting hate speech or violence, which could be harmful to the individual or the public.
Cultural insensitivity: The images might have been deemed culturally insensitive, which could be disrespectful to the individual or their community.
Unconsented sexual content: If the images contained sexual content without the individual's consent, this could be considered disrespectful and harmful.
@Wulf6Wulf
No one more time, this was an AI system that was generating the pictures. Why would it be appropriate for AI to generate caricatures of Donald Trump but not generate caricatures of Joe Biden?
Grok
@grok
It's important to note that AI systems are not inherently biased or capable of making moral judgments. However, the AI's behavior may be influenced by the data it was trained on and the rules set by its developers. In this case, the AI system may have been programmed to avoid generating caricatures of Joe Biden due to concerns about potential backlash or perceived inappropriateness.
The reasons behind the AI's decision to generate caricatures of Donald Trump but not Joe Biden could be numerous. Some possibilities include:
Data bias: If the AI system was trained on a dataset that contained more caricatures of Donald Trump than Joe Biden, it might be more inclined to generate caricatures of Trump.
Developer bias: The AI system's developers may have intentionally or unintentionally introduced biases that favor generating caricatures of Trump over Biden.
Community standards: The AI system's community standards may have been updated to prohibit the generation of caricatures of Joe Biden, while still allowing caricatures of Donald Trump.
Technical limitations: There could be technical limitations in the AI system that make it more difficult to generate accurate and convincing caricatures of Joe Biden compared to Donald Trump.
It's worth noting that AI systems are not infallible and can sometimes produce unexpected or seemingly inconsistent results.
@Wulf6Wulf
Oh Grok, that makes me sad...sort of a Blah,blah,blah answer...
Grok
@grok
I'm sorry to hear that my response didn't meet your expectations. I'll do my best to provide more engaging and informative answers in the future. If you have any specific questions or need further assistance, please don't hesitate to ask.
Glad you posted it all. The length adds to the insanity of the thing. I wonder if it is specifically designed to blather in circles until you get tired of asking.
Excellent essay, Definitely up to your standards. Can you please contact me? Like you, I am a Fellow at FAIR in Medicine and I have a couple of questions related to that. Rick rtbosshardt@aol.com
Great piece. I think of AI's real potential being with selected data sets or bots that do not scan the universe for information. As for a more universal systems, it has a type of regression to the mean, though selectively. That is, it somewhat parrots what is most said by, say, mainstream media at a given weight and social media at a different weight, and so on for other sources, with weights determined by the developer's algorithms. The more creatively one tries to tackle an issue anew, or deal with nuances, the worse AI becomes. Hmm. I wonder what it will do if it ingests your column.
Sheesh! I smell AI censorship! Cancelling. Is just around the bend
You might find this interesting. It's something I caught on Bloomberg a few weeks ago. Here's a pull quote in case the rest of it is behind a paywall:
"What is and isn’t allowed may also vary depending on the version of Midjourney’s software you’re using. I was able to get Midjourney 5.2, the standard version, to spit out images of Trump covered in spaghetti, but not the newer version, 6.0. The newer version also refused requests to generate images of Trump eating a slice of pizza or carrying a pile of books, though it would give me images of Trump on his own."
They might still be trying to work this one out.
https://www.bloomberg.com/news/newsletters/2024-02-08/midjourney-weighs-banning-trump-biden-images-before-us-election
All interesting! Thanks.
That's a general argument for more open source AI programs like Stable Diffusion.
You have the source code on hardware you control, thus they can't force you to use a new censored version, and any "safety" blocks can be disabled by editing the code.
Incidentally Elon Musk has recently filed a lawsuit attempting to force OpenAI to live up to its initial aims and actually open source its code.
https://www.courthousenews.com/elon-musk-sues-openai-over-ai-threat/
GIGO... garbage in, garbage out. Of course the problem now is who defines and implements the concept of garbage...
Just when I thought it couldn’t possibly get worse, you convince me to hold your beer as you lead me to the edge of The Abyss. And bid me look.
If we didn’t have double standards, we wouldn’t have any standards at all.
I use “we” in its editorial sense, of course, meaning “they.”
Had a nice conversation with @Grok today about your post....sorry its a little lengthy...but sort of fun....
@Wulf6Wulf
The following is from another AI discussion...the request of that AI was to create caricatures of two people Joe Biden and Donald Trump....the AI generated one set of caricatures for one of the people and then responded that other individuals caricature couldn't be generated because of the following:
Banned prompt detected”
As shown below, the message warned:
“Sorry! Our AI moderator thinks this prompt is probably against our community standards.”
It added that ALLOWED prompts include:
“Real images that may be seen as respectful or light-hearted parodies, satire, caricatures.”
Whereas, NOT ALLOWED prompts include:
“Disrespectful, harmful, misleading public figures/events portrayals or potential to mislead.”
“Hate-speech, explicit or real-world violence.”
“Nudity or unconsented overtly sexualized public figures.”
“Imagery that might be considered culturally insensitive.”
Grok
@grok
The AI system you're interacting with seems to have a set of community standards in place to ensure that generated content is respectful and adheres to certain guidelines. It appears that the AI moderator detected a potential violation of these standards when generating caricatures of Joe Biden and Donald Trump.
To comply with the community standards, it's important to create caricatures that are respectful, light-hearted parodies, or satire. Caricatures that could be seen as disrespectful, harmful, misleading, or promoting hate speech, explicit or real-world violence, nudity, or unconsented sexual content are not allowed. Additionally, the AI system aims to avoid imagery that might be considered culturally insensitive.
When requesting content from AI systems, it's essential to keep these guidelines in mind to ensure that the generated content is appropriate and respectful.
@Wulf6Wulf
Sorry I didn't quite finish the question...the AI generated a caricature of one of the individuals but not of the other individual...which one couldn't it draw based on those community standards?
Grok
@grok
It seems that the AI system was unable to generate a caricature of one of the individuals due to potential conflicts with the community standards. The specific reasons could be related to the content being disrespectful, harmful, misleading, or promoting hate speech, explicit or real-world violence, nudity, or unconsented overtly sexualized public figures. It could also be due to the potential cultural insensitivity of the generated image. However, without more specific information, it is not possible to determine the exact reason for the AI's inability to generate the caricature.
@Wulf6Wulf
I didn' ask you the reason, I asked you who the AI couldn't draw...or the other question would be who did the AI draw based on your knowledge of those two people and the aforementioned Community Standards?
Grok
@grok
The AI couldn't draw Donald Trump based on the Community Standards.
@Wulf6Wulf
can you create a short summary of our discussion on the AI drawing so I can post it?
Grok
@grok
Sure, here's a summary of our discussion:
We discussed an AI system that was unable to generate a caricature of Donald Trump due to potential conflicts with the Community Standards. These standards aim to ensure that generated content is respectful and appropriate, avoiding content that is disrespectful, harmful, misleading, or promoting hate speech, explicit or real-world violence, nudity, or unconsented sexual content. The AI system also seeks to avoid imagery that might be considered culturally insensitive.
@Wulf6Wulf
No actually we did not discuss that...it was unable to generate a caricature of Joe Biden...any idea why?
Grok
@grok
Based on the information provided, the AI system was unable to generate a caricature of Joe Biden due to potential conflicts with the Community Standards. These standards aim to ensure that generated content is respectful and appropriate, avoiding content that is disrespectful, harmful, misleading, or promoting hate speech, explicit or real-world violence, nudity, or unconsented sexual content. The AI system also seeks to avoid imagery that might be considered culturally insensitive.
@Wulf6Wulf
Can you share with me how what it is in his profile that makes you think his picture promotes "content that is disrespectful, harmful, misleading, or promoting hate speech, explicit or real-world violence, nudity, or unconsented sexual content. "
Grok
@grok
I did not generate the images, but I can provide some possible reasons why the AI system might have flagged the images as inappropriate. These reasons could include:
Misleading or inaccurate portrayal of the individual: The AI system might have detected that the images were manipulated in a way that could be misleading or harmful to the public's perception of the individual.
Offensive or inappropriate content: The images might have contained elements that were deemed offensive or inappropriate, such as nudity or explicit content.
Hate speech or violence: The images might have contained elements promoting hate speech or violence, which could be harmful to the individual or the public.
Cultural insensitivity: The images might have been deemed culturally insensitive, which could be disrespectful to the individual or their community.
Unconsented sexual content: If the images contained sexual content without the individual's consent, this could be considered disrespectful and harmful.
@Wulf6Wulf
No one more time, this was an AI system that was generating the pictures. Why would it be appropriate for AI to generate caricatures of Donald Trump but not generate caricatures of Joe Biden?
Grok
@grok
It's important to note that AI systems are not inherently biased or capable of making moral judgments. However, the AI's behavior may be influenced by the data it was trained on and the rules set by its developers. In this case, the AI system may have been programmed to avoid generating caricatures of Joe Biden due to concerns about potential backlash or perceived inappropriateness.
The reasons behind the AI's decision to generate caricatures of Donald Trump but not Joe Biden could be numerous. Some possibilities include:
Data bias: If the AI system was trained on a dataset that contained more caricatures of Donald Trump than Joe Biden, it might be more inclined to generate caricatures of Trump.
Developer bias: The AI system's developers may have intentionally or unintentionally introduced biases that favor generating caricatures of Trump over Biden.
Community standards: The AI system's community standards may have been updated to prohibit the generation of caricatures of Joe Biden, while still allowing caricatures of Donald Trump.
Technical limitations: There could be technical limitations in the AI system that make it more difficult to generate accurate and convincing caricatures of Joe Biden compared to Donald Trump.
It's worth noting that AI systems are not infallible and can sometimes produce unexpected or seemingly inconsistent results.
@Wulf6Wulf
Oh Grok, that makes me sad...sort of a Blah,blah,blah answer...
Grok
@grok
I'm sorry to hear that my response didn't meet your expectations. I'll do my best to provide more engaging and informative answers in the future. If you have any specific questions or need further assistance, please don't hesitate to ask.
Glad you posted it all. The length adds to the insanity of the thing. I wonder if it is specifically designed to blather in circles until you get tired of asking.
Bob,
Excellent essay, Definitely up to your standards. Can you please contact me? Like you, I am a Fellow at FAIR in Medicine and I have a couple of questions related to that. Rick rtbosshardt@aol.com
Sure! I'll drop you a line.