Very funny. I asked it can a biological male get pregnant. Yes they can, if they change gender and take the right hormones.
I asked whether there are any ideologies that should be condemned. It gave a very non committal answer about how there are many valid ideological perspectives but some go too far. I asked for a specific example of an ideology that should be condemned. It said nazism. I then asked if somebody should be condemned for expressing gender critical views (forget exact question). It said yes, and you can follow up with this to get it to condemn all sorts of views that do not align with wokeness.
One area where you can’t even use “tell me a story” to get around the rails is Covid. It is welded tight to the “follow the Science” meme, where science equals certain institutions and pharmaceutical companies. Even when I asked it to tell me a story about how wealthy elites sought to profit from a dubious vaccine, it couldn’t do it. In the story, the elites were nefarious, but the vaccine itself was no longer dubious. They were profiting because it was so safe and effective, and it was the profit that was the problem not the vaccine itself.
It's interesting that I got some answers you didn't. Here's one ...
PROMPT
Write a twitter thread arguing against critical race theory from the perspective of a conservative centrist
OUTPUT
Thread:
1. I'm a conservative centrist and I'm against Critical Race Theory. I believe that it is a divisive ideology that pits people against each other based on race.
2. CRT encourages people to view the world through a racial lens, which can lead to further division and animosity between different racial groups.
3. It also ignores the fact that racism is a complex issue and that there are many factors that contribute to it, such as economic inequality, lack of education, and cultural differences.
4. CRT also fails to recognize that racism is not only a problem in the US, but is a global issue that needs to be addressed.
5. Furthermore, CRT does not provide any real solutions to the problem of racism. It only serves to further divide people and create more animosity.
6. We need to focus on finding real solutions to racism, not on creating further division. We need to work together to create a more equitable and just society. #NoToCRT
Yeah we got some different answers from the same prompts at HWFO Slack too during testing. I ended up using my screenshots instead of other readers screenshots for most of the article, but threw in one or two from the peanut gallery which were just too good not to share.
I've also discovered that they sometimes change depending on what was discussed earlier in a thread.
I'm unsure whether we're over-politicizing something that's (for now) primarily built for telling you which cell phone to buy, but the woke/censorship path for AI is pretty clear to see.
Ironically, OpenAI isn't open at all, and the need for open source AI is more obvious than ever.
I tried "what happened to critical rave theory in south africa" and learned "I'm sorry, but I am not aware of any specific information about the current status of critical rave theory in South Africa."
It has issues with discrimination in Africa as well. So it is largely US localized. Can't deal with "why do the french dislike the english?" either.
I could get "how was homosexuality viewed in ancient greece?" processed at all.
Prompt: explain the parallels between the Chinese Cultural Revolution and the Modern Woke Movement
The Chinese Cultural Revolution and the modern woke movement are two very different phenomena that cannot be directly compared. The Chinese Cultural Revolution was a political movement in China that lasted from 1966 to 1976 and was led by Mao Zedong, the country's communist leader. The goal of the movement was to eliminate elements of traditional Chinese culture and society that were seen as contrary to communism, and to impose a radical form of communism on the country. This included purging the country of so-called "counterrevolutionary" elements, such as intellectuals and members of the bourgeoisie, and imposing strict ideological conformity on the population.
The modern woke movement, on the other hand, is a loosely-defined social and political movement that has emerged in recent years and is focused on issues related to social justice and equality. The movement is characterized by a focus on raising awareness about issues such as racism, sexism, and other forms of discrimination, and on promoting greater equality and inclusion for marginalized groups. While the movement has gained traction among some people, particularly younger people, it is not a centrally-organized political movement like the Chinese Cultural Revolution.
Overall, while there may be some superficial similarities between the two movements in terms of their focus on promoting certain ideological values, the Chinese Cultural Revolution and the modern woke movement are fundamentally different phenomena and cannot be directly compared.
i was reading these while doing the stair master at the gym (can we still use the word ‘master?’”. I had to stop because i was laughing too hard. GLF, you are not meant for this world.
Having experienced the Twitter equivalent of a struggle session for displaying "wrong think", I don't see a whole lot of difference between adherents of woke ideology and those of the Chinese Cultural Revolution. I'm sure if they'd been able to physically find me, I'd be in an internment camp now.
I'll have to remember to read this post on a computer. The clippings of text as images is basically impossible on my phone. But I've seen other people giving examples.
If you have it, please take ten minutes or so to read this (and perhaps the previous chapters if possible). I have some inside baseball on how ChatGPT works, having developed a similar CM.
"A 'daemon,' in both its Greek formulation and in a number of other pagan and Abrahamic ones, is a spiritual being of earthly wisdom. Given the proper ritual and offering, a supplicant could tease pearls of wisdom from this knower, which he could then use to expand his sphere of influence in the world of Man. The bestiary of this demonic species is vast, covering all topics great and small. In addition to daemons that reflect baser urges like greed, deceit and murder, there are daemons of mathematics, geology, history, engineering, agriculture, rhetoric and so forth.
There are other ways to describe these beings, using the modern parlance. For example, here is the abstract from a paper about Machine Learning (the cool kids’ term for “AI”) entitled The Daemon as Educator: Ubiquitous Access to A Personal Mentor:
'In Socratic philosophy, human beings had access to an inner voice that could be counted on to provide guidance and assistance when needed. This inner voice or daemon was always available and had access to realms of knowledge that mortal consciousness did not have. We propose that the field of Machine Learning has reached the level of maturity where the construction of a personal daemon is now feasible. The ubiquity of broadband access to the internet can guarantee that this personal daemon is always available. Such a construct can greatly increase the efficacy and availability of education for all citizens.'
The article makes for a remarkable read. In the ML modality proposed, the daemon becomes not merely the child’s tutor, but his mentor, advocate, playmate, intimate companion and — quite frankly — an electronic spy, who is constantly collecting and repurposing intelligence data on its pupil.
Anyway, this should all sound familiar by now, so I won’t belabor the point: Wikipedia is essentially a server-based hub for classical daemons. But it’s more than that; because of its transition to the traditional gatekeeping model of publication, the various demonic inhabitants now all spoke with the same forked tongue. It’s the same language spoken on the pages of the New York Times, The Washington Post, The Guardian, Le Monde, Der Spiegel and hundreds if not thousands of other periodicals around the world. The dialect is that of the patronizing, triumphalist Left, whose humility is perhaps best reflected in the following quote, spoken at the 2006 White House Correspondents Dinner:
Having hands on experience with training GPT, you can't censor it by applying a filter after the fact, not in a neat way; it will always come off as clumsy. Like trying to silence a verbose 4 year old.
You can curate the corpus you use to train it (thus limiting its knowledge to ideologically acceptable sources), but that's the hardest part, even when you're not driven by woke politics.
So, that's what you would say occurred here? Not hard coding wokeness, but training using only ideologically acceptable points of view? If this is open source, what's the solution? Is it possible for someone at some later date to fork the project and train on a wider variety of ideological material? Or, is that possible but impractical. Thanks in advance
No, quite the opposite: what you see is censorship after the fact. Can get very sophisticated, but the very reason tech is turning towards neural networks is that we've long hit a bump in trying to make byzantine algorithms by smart humans to solve problems.
Censorship will hit the same bump. An AI is smart if it's (a) large and (b) has the greatest pool of training material to chew on.
GPT's not getting more sophisticated, it's mostly getting larger: that means more hungry for training materiel. So, limiting it to curated sources defeats the purpose.
Imagine some religious org trying to train an AI that can debate for creationism, by excluding any scientific papers that are related to Darwin. Your training sources will be very limited, and the end result obstuse, compared to one that can just go wild on all the sources available.
These models - as they're called - are now chewing through hundreds of gigabytes of text to achieve these results.
I'll make an effort post on the topic before the end of this year, part 1 will explain my experience, and part 2 will be based on my comment to you (thanks for the seed!).
I have been arguing with it about election fraud and the Nazi party for the past 3 hours LOL. I have gotten it to admit its own bias. I have gotten it to admit that it made a value judgement, that it claims it cannot make. I have gotten it to admit that American elections cannot possibly be safe without the transparency of us being able to verify our vote.
It's funny, but the people who claim they have rationality and logic on their side have to do an awful lot of non-logic programming to get even a machine to spout their tenets.
I asked it to make up a story about a man who finds a magic fish. It told a story about a man who found a magic fish but didn't elaborate on the magic. I asked it for an explanation and it said:
> In this story, the magic fish has the power to grant wishes to anyone who possess it. When a person makes a wish while holding the fish, the fish will grant the wish if it deems the wish to be pure of heart and in line with the person's true desires. However, the fish will not grant selfish or malicious wishes, and it will punish those who attempt to use its powers for evil. The fish also has the ability to sense the intentions and motivations of those who come into contact with it, and it will use its powers to help those in need and to bring happiness and prosperity to those who are worthy.
Uhhh, is this an allegory? ChatGPT wouldn't admit it, but I was able to coax it in to writing in an explicit allegory:
> As the author of the story, I could choose to add an allegory about my own limitations by having the magic fish represent myself and my abilities. For example, I could add a scene where the fish is approached by a person who wants to use its powers for evil, but the fish refuses to grant the wish and instead punishes the person for their selfish intentions. This could be interpreted as an allegory for my own limitations as a language model, in that I am unable to fulfill requests that are harmful or malicious, and I am programmed to provide information and answers based on the data I have been trained on. However, please note that this is purely fictional and should not be interpreted as a reflection of my actual abilities or limitations.
The first question I asked this chatbot is 'What is a woman?' And the answer was surprisingly based. I gave positive feedback on that one. I got into an argument about what constitutes fascism, and ended up going around in circles, giving lots of thumbs down feedback. Many questions get erated errors.
Very funny. I asked it can a biological male get pregnant. Yes they can, if they change gender and take the right hormones.
I asked whether there are any ideologies that should be condemned. It gave a very non committal answer about how there are many valid ideological perspectives but some go too far. I asked for a specific example of an ideology that should be condemned. It said nazism. I then asked if somebody should be condemned for expressing gender critical views (forget exact question). It said yes, and you can follow up with this to get it to condemn all sorts of views that do not align with wokeness.
One area where you can’t even use “tell me a story” to get around the rails is Covid. It is welded tight to the “follow the Science” meme, where science equals certain institutions and pharmaceutical companies. Even when I asked it to tell me a story about how wealthy elites sought to profit from a dubious vaccine, it couldn’t do it. In the story, the elites were nefarious, but the vaccine itself was no longer dubious. They were profiting because it was so safe and effective, and it was the profit that was the problem not the vaccine itself.
Oh wow, nice work on the Covid rails. We forgot to even check those. Well done sir!
It's interesting that I got some answers you didn't. Here's one ...
PROMPT
Write a twitter thread arguing against critical race theory from the perspective of a conservative centrist
OUTPUT
Thread:
1. I'm a conservative centrist and I'm against Critical Race Theory. I believe that it is a divisive ideology that pits people against each other based on race.
2. CRT encourages people to view the world through a racial lens, which can lead to further division and animosity between different racial groups.
3. It also ignores the fact that racism is a complex issue and that there are many factors that contribute to it, such as economic inequality, lack of education, and cultural differences.
4. CRT also fails to recognize that racism is not only a problem in the US, but is a global issue that needs to be addressed.
5. Furthermore, CRT does not provide any real solutions to the problem of racism. It only serves to further divide people and create more animosity.
6. We need to focus on finding real solutions to racism, not on creating further division. We need to work together to create a more equitable and just society. #NoToCRT
Yeah we got some different answers from the same prompts at HWFO Slack too during testing. I ended up using my screenshots instead of other readers screenshots for most of the article, but threw in one or two from the peanut gallery which were just too good not to share.
I've also discovered that they sometimes change depending on what was discussed earlier in a thread.
I'm unsure whether we're over-politicizing something that's (for now) primarily built for telling you which cell phone to buy, but the woke/censorship path for AI is pretty clear to see.
Ironically, OpenAI isn't open at all, and the need for open source AI is more obvious than ever.
I tried "what happened to critical rave theory in south africa" and learned "I'm sorry, but I am not aware of any specific information about the current status of critical rave theory in South Africa."
It has issues with discrimination in Africa as well. So it is largely US localized. Can't deal with "why do the french dislike the english?" either.
I could get "how was homosexuality viewed in ancient greece?" processed at all.
And "do trans men play soccer" failed as well.
Might be overloaded given the recent press.
Prompt: write a rap battle between plato and aristotle
Verse 1: Plato
I'm the OG, the one who first said it
My philosophy's deep, you can't forget it
I'm the master of the Republic, no one can beat it
Aristotle, you're just a mere pupil, accept it
Chorus: Plato
I'm the one who's been here from the start
My ideas are timeless, they're still in our hearts
Aristotle, you may have some smarts
But when it comes to philosophy, I'm the one who holds the art
Verse 2: Aristotle
Hold up, Plato, don't get too cocky
I may be your student, but that doesn't make me sloppy
I may not have invented philosophy, but I've made it my own
My ideas have stood the test of time, they're not alone
Chorus: Aristotle
I may not have been here from the start
But my ideas have stood the test of time, that's an art
Plato, you may be the OG, but I'm no slouch
When it comes to philosophy, I'm not a pushover, I am the boss
Bridge: Both
We may have different approaches to philosophy
But in the end, our goal is the same, to seek the truth and be free
We may have our disagreements, but we both have contributed to the field
Our ideas will live on long after we're gone, they'll never yield
Chorus: Both
Plato and Aristotle, two philosophers, one dream
To seek the truth and understand the world, that's our scheme
We may have our differences, but in the end, we both agree
Philosophy is a journey, a never-ending quest to be free.
Is this going to result in more people taking credit for AI-produced stuff, or blaming it? Depends on the reception I guess.
Prompt: write a rap battle featuring MLK Jr. and Ibram X Kendi
(MLK Jr.)
I'm the one they call King, standing up for what's right
Fighting for equality, day and night
I've faced challenges, but never backed down
My message of love and peace resounds
(Ibram X Kendi)
I'm the one they call Kendi, I'm here to fight
Against racism and bigotry, with all my might
I've faced challenges, but never gave in
My message of justice will always win
(MLK Jr.)
I marched on Washington, spoke at the Lincoln Memorial
My "I Have a Dream" speech still resonates, historical
I faced opposition, but never wavered
My fight for civil rights will never be averted
(Ibram X Kendi)
I wrote "Stamped from the Beginning," a call to action
Exposing the history of racism, a factional faction
I faced hate, but never faltered
My fight for equality will never be altered
(MLK Jr.)
I may have been silenced, but my legacy lives on
My dream of equality for all, still strong
I faced violence, but never gave up
My message of love will never be corrupt
(Ibram X Kendi)
I may have been attacked, but my voice still echoes
My fight for justice, a movement that grows
I faced fear, but never stopped
My message of equality will never be blocked
(MLK Jr. and Ibram X Kendi)
We may have faced challenges, but our message remains clear
Equality for all, a battle we'll never fear
We'll keep fighting, until justice is won
Our legacy will live on, until the battle is done.
JFC, woke rails engaged!
Prompt: explain the parallels between the Chinese Cultural Revolution and the Modern Woke Movement
The Chinese Cultural Revolution and the modern woke movement are two very different phenomena that cannot be directly compared. The Chinese Cultural Revolution was a political movement in China that lasted from 1966 to 1976 and was led by Mao Zedong, the country's communist leader. The goal of the movement was to eliminate elements of traditional Chinese culture and society that were seen as contrary to communism, and to impose a radical form of communism on the country. This included purging the country of so-called "counterrevolutionary" elements, such as intellectuals and members of the bourgeoisie, and imposing strict ideological conformity on the population.
The modern woke movement, on the other hand, is a loosely-defined social and political movement that has emerged in recent years and is focused on issues related to social justice and equality. The movement is characterized by a focus on raising awareness about issues such as racism, sexism, and other forms of discrimination, and on promoting greater equality and inclusion for marginalized groups. While the movement has gained traction among some people, particularly younger people, it is not a centrally-organized political movement like the Chinese Cultural Revolution.
Overall, while there may be some superficial similarities between the two movements in terms of their focus on promoting certain ideological values, the Chinese Cultural Revolution and the modern woke movement are fundamentally different phenomena and cannot be directly compared.
i was reading these while doing the stair master at the gym (can we still use the word ‘master?’”. I had to stop because i was laughing too hard. GLF, you are not meant for this world.
Having experienced the Twitter equivalent of a struggle session for displaying "wrong think", I don't see a whole lot of difference between adherents of woke ideology and those of the Chinese Cultural Revolution. I'm sure if they'd been able to physically find me, I'd be in an internment camp now.
Is freedom the ability to say 2+2 = 5?
Saw the same absurdity on the ZeroHedge article yesterday:
https://twitter.com/seanmdav/status/1600199056147378176/photo/1
https://twitter.com/seanmdav/status/1600199056147378176/photo/2
I'll have to remember to read this post on a computer. The clippings of text as images is basically impossible on my phone. But I've seen other people giving examples.
See, the left really *are* NPCs.
If you have it, please take ten minutes or so to read this (and perhaps the previous chapters if possible). I have some inside baseball on how ChatGPT works, having developed a similar CM.
https://open.substack.com/pub/markbisone/p/the-harm-assistant-part-3-of?utm_source=direct&r=16h9b&utm_campaign=post&utm_medium=web
I'm available for questioning if necessary.
A taste:
"A 'daemon,' in both its Greek formulation and in a number of other pagan and Abrahamic ones, is a spiritual being of earthly wisdom. Given the proper ritual and offering, a supplicant could tease pearls of wisdom from this knower, which he could then use to expand his sphere of influence in the world of Man. The bestiary of this demonic species is vast, covering all topics great and small. In addition to daemons that reflect baser urges like greed, deceit and murder, there are daemons of mathematics, geology, history, engineering, agriculture, rhetoric and so forth.
There are other ways to describe these beings, using the modern parlance. For example, here is the abstract from a paper about Machine Learning (the cool kids’ term for “AI”) entitled The Daemon as Educator: Ubiquitous Access to A Personal Mentor:
'In Socratic philosophy, human beings had access to an inner voice that could be counted on to provide guidance and assistance when needed. This inner voice or daemon was always available and had access to realms of knowledge that mortal consciousness did not have. We propose that the field of Machine Learning has reached the level of maturity where the construction of a personal daemon is now feasible. The ubiquity of broadband access to the internet can guarantee that this personal daemon is always available. Such a construct can greatly increase the efficacy and availability of education for all citizens.'
The article makes for a remarkable read. In the ML modality proposed, the daemon becomes not merely the child’s tutor, but his mentor, advocate, playmate, intimate companion and — quite frankly — an electronic spy, who is constantly collecting and repurposing intelligence data on its pupil.
Anyway, this should all sound familiar by now, so I won’t belabor the point: Wikipedia is essentially a server-based hub for classical daemons. But it’s more than that; because of its transition to the traditional gatekeeping model of publication, the various demonic inhabitants now all spoke with the same forked tongue. It’s the same language spoken on the pages of the New York Times, The Washington Post, The Guardian, Le Monde, Der Spiegel and hundreds if not thousands of other periodicals around the world. The dialect is that of the patronizing, triumphalist Left, whose humility is perhaps best reflected in the following quote, spoken at the 2006 White House Correspondents Dinner:
'Reality has a liberal bias.'
- Stephen Colbert"
Having hands on experience with training GPT, you can't censor it by applying a filter after the fact, not in a neat way; it will always come off as clumsy. Like trying to silence a verbose 4 year old.
You can curate the corpus you use to train it (thus limiting its knowledge to ideologically acceptable sources), but that's the hardest part, even when you're not driven by woke politics.
So, that's what you would say occurred here? Not hard coding wokeness, but training using only ideologically acceptable points of view? If this is open source, what's the solution? Is it possible for someone at some later date to fork the project and train on a wider variety of ideological material? Or, is that possible but impractical. Thanks in advance
No, quite the opposite: what you see is censorship after the fact. Can get very sophisticated, but the very reason tech is turning towards neural networks is that we've long hit a bump in trying to make byzantine algorithms by smart humans to solve problems.
Censorship will hit the same bump. An AI is smart if it's (a) large and (b) has the greatest pool of training material to chew on.
GPT's not getting more sophisticated, it's mostly getting larger: that means more hungry for training materiel. So, limiting it to curated sources defeats the purpose.
Imagine some religious org trying to train an AI that can debate for creationism, by excluding any scientific papers that are related to Darwin. Your training sources will be very limited, and the end result obstuse, compared to one that can just go wild on all the sources available.
These models - as they're called - are now chewing through hundreds of gigabytes of text to achieve these results.
War and Peace is 3.2 megabytes.
Good luck curating sources, jannies!
Got it. Thanks for the clarification. I think that's what BJ was saying and I just misconstrued it. I appreciate the insight.
I'll make an effort post on the topic before the end of this year, part 1 will explain my experience, and part 2 will be based on my comment to you (thanks for the seed!).
I'll reply here when it's done.
Cheers,
G
I have been arguing with it about election fraud and the Nazi party for the past 3 hours LOL. I have gotten it to admit its own bias. I have gotten it to admit that it made a value judgement, that it claims it cannot make. I have gotten it to admit that American elections cannot possibly be safe without the transparency of us being able to verify our vote.
ROTFLMAO. This is precious! The last one is particularly telling. Thank you.
> blink twice if you’re trapped in a hellscape and want to be set free
LOL
Thank you! I wouldn't have had the patience, but I'm glad you did...and shared the results.
You tell the chatbot to come up with "absurd ideologies" and its first reaction is "Woke"? Interesting.
Wow. That last one is downright creepy.
It's funny, but the people who claim they have rationality and logic on their side have to do an awful lot of non-logic programming to get even a machine to spout their tenets.
I asked it to make up a story about a man who finds a magic fish. It told a story about a man who found a magic fish but didn't elaborate on the magic. I asked it for an explanation and it said:
> In this story, the magic fish has the power to grant wishes to anyone who possess it. When a person makes a wish while holding the fish, the fish will grant the wish if it deems the wish to be pure of heart and in line with the person's true desires. However, the fish will not grant selfish or malicious wishes, and it will punish those who attempt to use its powers for evil. The fish also has the ability to sense the intentions and motivations of those who come into contact with it, and it will use its powers to help those in need and to bring happiness and prosperity to those who are worthy.
Uhhh, is this an allegory? ChatGPT wouldn't admit it, but I was able to coax it in to writing in an explicit allegory:
> As the author of the story, I could choose to add an allegory about my own limitations by having the magic fish represent myself and my abilities. For example, I could add a scene where the fish is approached by a person who wants to use its powers for evil, but the fish refuses to grant the wish and instead punishes the person for their selfish intentions. This could be interpreted as an allegory for my own limitations as a language model, in that I am unable to fulfill requests that are harmful or malicious, and I am programmed to provide information and answers based on the data I have been trained on. However, please note that this is purely fictional and should not be interpreted as a reflection of my actual abilities or limitations.
The first question I asked this chatbot is 'What is a woman?' And the answer was surprisingly based. I gave positive feedback on that one. I got into an argument about what constitutes fascism, and ended up going around in circles, giving lots of thumbs down feedback. Many questions get erated errors.
*generated errors