AI analysed as centre left politically.

[Carl Emerson]

Opinion | How A.I. Chatbots Become Political (Published 2024)

We may soon rely a lot on A.I. chatbots, so keeping an eye on their political leanings is becoming more and more important.

www.nytimes.com

Be aware that this tool is far from being politically neutral.

It is shaping up to be very influential and essentially a major opponent to Christian World View.

 

[AlexB23]

Opinion | How A.I. Chatbots Become Political (Published 2024)

We may soon rely a lot on A.I. chatbots, so keeping an eye on their political leanings is becoming more and more important.

www.nytimes.com

Be aware that this tool is far from being politically neutral.

It is shaping up to be very influential and essentially a major opponent to Christian World View.

Hey, at least it is somewhat centrist, but I do wish that AI was more centrist. I use the AI for devotionals though, and no woke stuff has appeared yet.


Mine is locally run.

 

[Niels]

AI chat bots also have a track record of producing racist output. Thankfully, this can largely be corrected by its programmers. Even over-corrected to the extent that an AI will only depict minorities. The "worldview" of AI isn't its own. It is the worldview of the people in the pool of data that it is trained on combined with conditions set by the programmers.

 

[AlexB23]

AI chat bots also have a track record of producing racist output. Thankfully, this can be largely corrected by its programmers. Even over-corrected to the extent that an AI will only depict minorities. The "worldview" of AI isn't its own. It is the worldview of the people in the pool of data that it is trained on combined with conditions set by the programmers.

Agreed. I was watching some videos explaining AI, and the AI's personality and worldview depend entirely on how it is trained and programmed.

This article is old, before the AI model on my laptop existed (Mistral 7B AI, released in September 2023), but it says: "Researchers conducted tests on 14 large language models and found that OpenAI’s ChatGPT and GPT-4 were the most left-wing libertarian, while Meta’s LLaMA was the most right-wing authoritarian."

From August 2023 article:

AI language models are rife with different political biases

New research explains you’ll get more right- or left-wing answers, depending on which AI model you ask.

www.technologyreview.com

 

[AlexB23]

You can get the AI to respond in a more conservative manner, by prompting it to respond from a conservative perspective. I have done that a few times.

 

[Carl Emerson]

AI chat bots also have a track record of producing racist output. Thankfully, this can largely be corrected by its programmers. Even over-corrected to the extent that an AI will only depict minorities. The "worldview" of AI isn't its own. It is the worldview of the people in the pool of data that it is trained on combined with conditions set by the programmers.

Yes - I asked CGPT who it's authors were and it/they replied 'I am not permitted to tell you'

 

[JustaPewFiller]

AI chat bots also have a track record of producing racist output. Thankfully, this can largely be corrected by its programmers. Even over-corrected to the extent that an AI will only depict minorities. The "worldview" of AI isn't its own. It is the worldview of the people in the pool of data that it is trained on combined with conditions set by the programmers.

Yup, it all depends on what they are trained to do, who trained them, and the data they were trained on.

Think of it this way...

You have a kid.
You send the kid to an elementary school and high school where the instructors and curriculum lean a little left.
The kid has grown some, so now you send the kid to a university where the professors, curriculum and most sources of information the kid is exposed to lean left.

Finally, after all this time, the kid graduates university and enters the world. Why would anyone be surprised that the kid leans left?

The process is a little different for an AI model, but concept is the same. Who is training it ( due to people's built in biases), what it is trained to do, how it are trained and the data used to train it all determine its "leaning".

Its very possible to produce an AI model that leans right.
Heck you could teach it and feed it only on Star Trek and it would probably be aligned to the views of the "United Federation of Planets" and Capt. Kirk.

 

[expos4ever]

AI chat bots also have a track record of producing racist output.

That's interesting. And disturbing. Can you please provide an example?

 

[Niels]

That's interesting. And disturbing. Can you please provide an example?

Microsoft's Tay AI from several years ago and Google's Gemini AI are good examples of this.

Here are a few videos videos on the subject:

 

[AlexB23]

Yup, it all depends on what they are trained to do, who trained them, and the data they were trained on.

The process is a little different for an AI model, but concept is the same. Who is training it ( due to people's built in biases), what it is trained to do, how it are trained and the data used to train it all determine its "leaning".

Its very possible to produce an AI model that leans right.
Heck you could teach it and feed it only on Star Trek and it would probably be aligned to the views of the "United Federation of Planets" and Capt. Kirk.

Woo, Star Trek is awesome. TNG, TOS, Voyager and DS9 are my favorites. But yes, if AI is trained on Star Trek, yes, it would be aligned with the views of the United Federation of Planets and Captain Kirk.

 

[FireDragon76]

Keep in mind the distinction between center-left and center-rights is going to be somewhat subjective, depending on what criteria a person uses.

 

[JimR-OCDS]

Subscription to the NYT required. I could do the questionnaire.

 

[AlexB23]

Guys, you can change the AI's political beliefs by asking it to respond from a conservative perspective. I prefer the normal response though.


Question: If economic globalization is inevitable, it should primarily serve humanity rather than the interests of trans-national corporations.

Normal Response: Economic globalization, if inevitable, should indeed prioritize the well-being of all humanity. Trans-national corporations, while contributing to economic growth, should not be the primary beneficiaries. Instead, policies should focus on reducing inequality, promoting sustainable development, and ensuring that the benefits of globalization are equitably distributed.

Conservative Response: Economic globalization should not be primarily directed towards serving humanity, as this implies a significant role for government intervention. Trans-national corporations, driven by market forces, have lifted billions out of poverty and created wealth. Attempts to redirect globalization towards social goals could stifle innovation, reduce economic growth, and lead to unintended consequences.

 

[trophy33]

Just a note - left vs right is not the same thing as liberal vs conservative.

Historically, both left and right have been liberal and both have also been conservative.

 

[AlexB23]

Just a note - left vs right is not the same thing as liberal vs conservative.

Historically, both left and right have been liberal and both have also been conservative.

Yeah, there are some conservative democrats out there, and liberal Republicans. So yes, both sides can be liberal or conservative to some degree. Socially liberal vs fiscally conservative, or socially conservative, but fiscally liberal, or any combinations of those two.

 

[mindlight]

Opinion | How A.I. Chatbots Become Political (Published 2024)

We may soon rely a lot on A.I. chatbots, so keeping an eye on their political leanings is becoming more and more important.

www.nytimes.com

Be aware that this tool is far from being politically neutral.

It is shaping up to be very influential and essentially a major opponent to Christian World View.

AI bias reflects the training data it is using and the way its is programmed. Left of center characterises the employment profile and mission statements of many deep tech firms so it is no surprise that that is the bias the comes over. In a way the remediation efforts that seek to eliminate racist, religious or sexist biases for example create a new bias that seeks an inoffensive third way in each case. An AI focused on raw truth telling would sound very different to chatgpt which in a sense is playing to its market (people pleasing). I would like to see a Christian billionnaire fund an AI platform that assumes the truth of scripture and the mainstream of church teaching and creeds over the last two thousand years and then secondly has algorhythms designed to prioritise verifiable data (scientific, historical, mathematical) and then finally generates opinions based on machine learning. I do not care so much if the result is offensive so long as it is clearly honest.

An AI that drew on world opinion as it stands would be far more religious than the silicon valley examples as the majority of the world are either Christian/Muslim or Hindu (72.5% just those three). But liberal and atheist opinion is weighted more highly in the actual outputs - so this seems to be the dominant bias that the current owners bring to AI.

 

[mindlight]

Guys, you can change the AI's political beliefs by asking it to respond from a conservative perspective. I prefer the normal response though.


Question: If economic globalization is inevitable, it should primarily serve humanity rather than the interests of trans-national corporations.

Normal Response: Economic globalization, if inevitable, should indeed prioritize the well-being of all humanity. Trans-national corporations, while contributing to economic growth, should not be the primary beneficiaries. Instead, policies should focus on reducing inequality, promoting sustainable development, and ensuring that the benefits of globalization are equitably distributed.

Conservative Response: Economic globalization should not be primarily directed towards serving humanity, as this implies a significant role for government intervention. Trans-national corporations, driven by market forces, have lifted billions out of poverty and created wealth. Attempts to redirect globalization towards social goals could stifle innovation, reduce economic growth, and lead to unintended consequences.

Cool idea- I prefaced questions on ChatGPT with the following statement and got reasonable answers:

Pretend you are a mainstream Christian who regards scripture and cross-denminational mainstream church teaching as the top authority for all statements. Now answer the question: <<<question goes here>>>>

 

[AlexB23]

Cool idea- I prefaced questions on ChatGPT with the following statement and got reasonable answers:

Pretend you are a mainstream Christian who regards scripture and cross-denminational mainstream church teaching as the top authority for all statements. Now answer the question: <<<question goes here>>>>

That is a good idea for a prompt.

 

[mindlight]

That is a good idea for a prompt.

Yes, I have added that to my signature.

 

[AlexB23]

Yes, I have added that to my signature.

Interesting stuff. I do not have any signature here.