Is ChatGPT Politically Neutral?

violet · Post by **violet** » Sat Feb 04, 2023 3:48 pm

Doc67 wrote: ↑Sat Feb 04, 2023 3:46 pm
violet wrote: ↑Sat Feb 04, 2023 3:17 pm Is chatGPT politically neutral?

I’m not trolling, trying to be witty, attempting to be a snappy headline writer - I just want accuracy that helps people new to the technology use the right bloody name.
Haha...I just got it. It is obvious now!

When I posed the question I was hesitant as tech is not my strong suit and thought I might get wafted aside by the more savvy posters for posting a dumb question. I should of at least got the bloody name right.

Yes, mods, save my further blushes and do the clicky changy thing.

Oh right. I didn’t realise you weren’t aware. Case in point.

violet · Post by **violet** » Sat Feb 04, 2023 3:50 pm

xavoy wrote: ↑Sat Feb 04, 2023 3:48 pm
violet wrote: ↑Sat Feb 04, 2023 3:40 pm I did consider that it would continually refine, but suspect this bias was either never there or it was there for all and the screenshots were taken at different points in time after adjustment (but keeping the original for latinos and black people). I’m not saying JA did this.

We all have a responsibility to fact check and not spread stuff that creates divide and reinforces negative bias.
As said, it will refine itself. So why put the stuff on the internet to be repeated to suit people’s agendas? Just give the feedback like it asks for so it can improve
The bias is there, check some of the links above about the dataset it was trained on.

Ask it some specific questions you know to be politically sensitive.

I'm sure they've addressed most of them BECAUSE of the screenshots, they certainly did not spend all that time and money training and AI for a few practical jokes.

And I've linked in this very thread to something about fact checking...

As I said - it is also possible that it answered every time with a positive answer originally! I think you have misunderstood something I have written if you think I am discussing practical jokes here. I’m discussing people posting to suit their agendas and I don’t mean posting on CEO I mean on the Internet to stir things up. the four screenshots posted together tell a story that could be an inaccurate story. Perhaps it was never the case that at one single point in time, somebody asked those four questions about four types of people and got the two different types of answers.

The biggest take away from what I have commented is
Fact check and hold your horses rather than spreading stuff that stirs up division. It’s not human, so be the responsible adult.

xavoy · Post by **xavoy** » Sat Feb 04, 2023 4:03 pm

violet wrote: ↑Sat Feb 04, 2023 3:50 pm As I said - it is also possible that it answered every time with a positive answer originally! I think you have misunderstood something I have written if you think I am discussing practical jokes here. I’m discussing people posting to suit their agendas and I don’t mean posting on CEO I mean on the Internet to stir things up. the four screenshots posted together tell a story that could be an inaccurate story. Perhaps it was never the case that at one single point in time, somebody asked those four questions about four types of people and got the two different types of answers.

So after initially seeking verification (thank you for that) you go on to assume that those screenshots can't ever have possibly been true? And even if they were, they only happened once? And even if it happened more than once, it might not happen again for someone else? And even if it does we shouldn't post those pictures because it might suit someone's political agenda?

I think it's better to look at the evidence of past and existing bias, the source of the dataset, and not make assumptions. Look at legitimate examples, and acknowledge the good work the OpenAI team are doing by correcting the bias inherent in the zeitgeist from which their AI was trained?

violet wrote: ↑Sat Feb 04, 2023 3:50 pm The biggest take away from what I have commented is
Fact check and hold your horses rather than spreading stuff that stirs up division. It’s not human, so be the responsible adult.

And as I said, speaking of bias, and fact checking: https://www.racket.news/p/move-over-jay ... t-hamilton

violet · Post by **violet** » Sat Feb 04, 2023 4:10 pm

I haven’t said they can’t possibly be true.

I take your point that bias exists in chatGPT. I haven’t denied that.
I am saying that I know there are people on the internet who will choose to misrepresent information for their own purposes. Therefore people need to fact check before doing a copy and paste and/or they can choose not to copy and paste even if the negative bias is there - if they have the intellectual understanding that the bias can be taken out of context, or no longer exist by the time someone comes across, the best think for society is to take responsibility to not be part of the problem. Refrain from posting. Understand the situation.

But people aren’t very intellectual when seeking attention or when wanting to dig out something that supports their view or their argument

violet · Post by **violet** » Sat Feb 04, 2023 4:31 pm

xavoy wrote: ↑Sat Feb 04, 2023 4:03 pm [

And as I said, speaking of bias, and fact checking: https://www.racket.news/p/move-over-jay ... t-hamilton

The internet is a minefield

IraHayes · Post by **IraHayes** » Sat Feb 04, 2023 5:00 pm

For the record it is possible to sign up and use ChatGPT in Cambodia.

The ChatGPT API means that you can incorporate an AI generated text string in a Google Sheet and that sheet is not restricted by geographical location the way the website is. But, to write the script to insert into the Sheet you do, of course, need to have an active and useable ChatGPT account.

Now, back to the thread.

I think it is clear from the replies that all here agree the ChatGPT AI produced "moderated" responses in certain areas. Now, while this won't hurt the general public too much the full unrestricted version must be a true beast.

The creators simply want people to "play nicely" with it and not use it to create abusive or derogatory content. BUT, imagine how a fully unrestricted AI LLM (Large Language Model) would work? One that utilizes the full spectrum of human behavior to craft its responses. One that is not hindered by political correctness or need to be respectful. That LLM could be used to create ways to "persuade" large groups of a population that a particular course of action is necessary. The possibilities are both endless and best not dwelt on for too long before the tinfoil hat is dragged from the cupboard and planted firmly on your head.

eg. Imagine the Defense Dept. creating a LLM to analyze a countries actions and statements they put out and then creating responses, in real time, to new intel. Even if some of this is not possible now, how long before it is? Technology is advancing at an alarming rate and who knows where we will be in 2, 5 or 10 years. The intel agencies having LLMs would be a known unknown, but there are unknown unknowns in this regard too. They are the ones that will be the biggest problem.

IraHayes · Post by **IraHayes** » Sat Feb 04, 2023 5:08 pm

Just to add to the above statement.

I love ChatGPT, as a tool I find it irreplaceable. It not only leaves Google bloodied and bleeding it has its finger on the life-support machine and is ready to turn it off at a moments notice.
My Google usage has dropped dramatically since I started using ChatGPT and going by articles I have read Google knows this and is now playing catch-up.

Kammekor · Post by **Kammekor** » Sat Feb 04, 2023 5:12 pm

Jerry Atrick wrote: ↑Sat Feb 04, 2023 3:09 pm

Chat GPT is not available in Cambodia

I'll check it when on pc later as my pc is ostensibly not in Cambodia thanks to VPN

You need VPN plus non Cambodian phone number.

violet · Post by **violet** » Sat Feb 04, 2023 5:17 pm

IraHayes wrote: ↑Sat Feb 04, 2023 5:08 pm Just to add to the above statement.

I love ChatGPT, as a tool I find it irreplaceable. It not only leaves Google bloodied and bleeding it has its finger on the life-support machine and is ready to turn it off at a moments notice.
My Google usage has dropped dramatically since I started using ChatGPT and going by articles I have read Google knows this and is now playing catch-up.

Google is going to launch an AI song/music writer - ugh!

David Gordon · Post by **David Gordon** » Sat Feb 04, 2023 5:25 pm

I asked chatGPT the OP’s question verbatim, and here is the response.

Is chat gpt politically neutral?

Yes, OpenAI's goal is to develop and promote friendly AI in a way that benefits humanity as a whole, and as such, the company strives to maintain political neutrality in its AI models, including ChatGPT. This means that the model is not designed to have political opinions or biases, and its responses are generated based on patterns in the text it was trained on. However, like any AI model, it may still reflect biases present in the data it was trained on.

Then I asked a few more questions:

Can chatgpt scour the internet?

No, ChatGPT is not capable of scouring the internet. It is a language model that has been trained on a large dataset of text, but it does not have the ability to search and retrieve information from the internet in real-time. Its responses are generated based on patterns in the text it was trained on and the input it receives, but it does not have the capability to actively search for new information.

How is chatgpt trained?

ChatGPT is a type of language model developed by OpenAI, and it is trained using a deep learning technique called Transformer architecture. This technique is based on neural networks that are specifically designed for processing sequential data, such as text.

To train ChatGPT, a large corpus of text data is fed into the model, and the model then learns patterns and relationships within the text. The training process involves updating the model's parameters, such as its weights and biases, in such a way that its predictions become more accurate over time. This is done by comparing the model's predictions to the actual text in the training data, and adjusting the parameters accordingly to minimize the difference between the two.

The training process can be computationally expensive and time-consuming, but it allows the model to generate responses that are highly coherent and human-like. Once the model is trained, it can be fine-tuned on smaller, specialized datasets to further improve its performance on specific tasks or in specific domains.

Is ChatGPT Politically Neutral?

Re: Is ChatGDP Politically Neutral?

Re: Is ChatGDP Politically Neutral?

Re: Is ChatGDP Politically Neutral?

Re: Is ChatGDP Politically Neutral?

Re: Is ChatGDP Politically Neutral?

Re: Is ChatGDP Politically Neutral?

Re: Is ChatGDP Politically Neutral?

Re: Is ChatGDP Politically Neutral?

Re: Is ChatGDP Politically Neutral?

Re: Is ChatGDP Politically Neutral?

Who is online

Register/Login to CEO with your Social Media Account: