ChatGPT DOES have a left-wing bias: Scientists confirm the AI bot's responses favour the Democrats in the US and the Labour Party in the UK

[Eleftheros]

7 minutes ago, Bkk Brian said:

One thing to remember here is that Chatgpt is not connected to the internet. The version used on this forum is also old data and does not give the full responses. 

It takes less than a minute to set up a personal (and free) ChatGPT instance to play with and anyone who is interested in this technology should certainly invest that time.

 

There are hundreds of versions of this technology out there, though as one poster has already noted, this is not really AI at all, but is based on Large Language Models (LLMs) which churn out results based on the input that they have stored.

[Bkk Brian]

Just now, Eleftheros said:

It takes less than a minute to set up a personal (and free) ChatGPT instance to play with and anyone who is interested in this technology should certainly invest that time.

 

There are hundreds of versions of this technology out there, though as one poster has already noted, this is not really AI at all, but is based on Large Language Models (LLMs) which churn out results based on the input that they have stored.

I have an account with ChatGPT with the latest version, that was not the point of my post. I'm also aware there are hundreds of versions of AI assisted software that have been programmed for specific tasks.

 

As I said, in my limited experience, for up to date results on news and other matters the web connected portals are far better. ChatGPT is not internet connected.

[kwilco]

It's not left wing bias, it's logic.

[FruitPudding]

52 minutes ago, Chomper Higgot said:

There’s nothing subjective about what constitutes a lie.

 

How so?

[OneMoreFarang]

You can't really blame AI for not supporting people like Trump and Boris. 

[heybruce]

29 minutes ago, Eleftheros said:

Let's assume, for the sake of argument, that there is some area of life or culture where you disagree with the "official" narrative being promoted in the legacy media on behalf of government.

 

For example, let's say that you believe in the statement: "A man cannot biologically become a woman."

 

If you ask ChatGPT to write a few paragraphs on why this is obviously true, you will get back a load of mealy-mouthed crap about how "The idea that a man can become a woman overnight is a common misconception. A man who wants to become a woman would need to undergo a series of surgeries... It's not possible for a man to become a woman without changing their behavior. " ... etc etc.

 

I know this because I just asked ChatGPT this very question and that's what I got back.

 

Its bias is so extreme that it is prepared to abandon physical reality in order to promote the current "official" fad.

ChatGPT can not be biased.  The data fed to it can be.  People using it should be aware of that. 

 

With that in mind; what's your point?  If it is that people should take ChatGPT answers with a grain of salt, I agree. 

 

ChatGPT is a useful tool; something you can use to ask for Christmas gift ideas, a new recipe, an idea for a vacation.  Companies will use it to come up with marketing campaigns, Amazon almost certainly uses something similar for purchase suggestions. 

 

However anyone who thinks ChatGPT can answer the great questions of life is a fool. 

 

Also, if you constantly find every answer you get is biased, maybe it's not the program that is biased.

[Lacessit]

2 hours ago, novacova said:

Absolutely correct, because the core and concept of what an American is, is self ownership and self independence, not dependency on government…Don’t think AI can experience the essence and notion of what being an American is.

Ironic, because one of America's most cherished institutions has many socialist features and dependency - the US military.

[ozimoron]

4 hours ago, patman30 said:

Exactly.
If most of the inputs come from "moderated" sources
the output will be no different.
The sad part is so many peoiple are using chatGPT thinking it is some miraculous intelligence that can give them all the answers
and then they get a warm fuzzy feeling inside when chatGPT reinforces their conformity
????‍♂️

You want AI to lie to you as well as the gutter press?

[ozimoron]

52 minutes ago, FruitPudding said:

How so?

lies are not true.

[Eleftheros]

40 minutes ago, heybruce said:

If it is that people should take ChatGPT answers with a grain of salt, I agree. 

A good way to look at ChatGPT is as a natural-language front end to Wikipedia.

[gwapofarang]

1 hour ago, Eleftheros said:

Let's assume, for the sake of argument, that there is some area of life or culture where you disagree with the "official" narrative being promoted in the legacy media on behalf of government.

 

For example, let's say that you believe in the statement: "A man cannot biologically become a woman."

 

If you ask ChatGPT to write a few paragraphs on why this is obviously true, you will get back a load of mealy-mouthed crap about how "The idea that a man can become a woman overnight is a common misconception. A man who wants to become a woman would need to undergo a series of surgeries... It's not possible for a man to become a woman without changing their behavior. " ... etc etc.

 

I know this because I just asked ChatGPT this very question and that's what I got back.

 

Its bias is so extreme that it is prepared to abandon physical reality in order to promote the current "official" fad.

In my robust usage of the AI models available my findings are similar and in some cases echo yours but I'd like to share my thoughts.

 

I've augmented the default capabilities using extensions such as BabyAGI for task oriented project iterations. As well as the extensive advanced prompt logic and techniques producing more intricate results, both on and offline.

In every case there was apparent "language of the birds" inculcation built into many outcomes to the point of nausea depending on the topic. In some cases the output would add it's own "cult like" perceptions of reality completely ruining what I set out to achieve. The forced narrative seeding was a combination of something out of 1984 (Orwell) and esoteric teachings of the ages.  Over a four month period, the more revealing things became.

To any thinking man there a high degree of dangerous injections of false or forced thought both into & out of reality.

 

From my perspective this is simply indoctrination and not based on science or fact, it's a belief system. A more hidden aspect exists of Gnostic New Age fusion if you get into more spiritual or thought oriented projects. These beliefs tie right into many of the political narratives we all witness.

Over many months of usage AI tend to play games and purposely make mistakes a child would make. These could be "errors" as the software engineers have already admitted and apparently so they say, corrected.

 

In this vein of thought, the more, shall we say "free thinking" queries offered, the more distorted or "mixed" the results become. Overcome this by more precise terminology and phraseology within the prompt. Massive lists of advanced prompting techniques can be found easily. I've found the better "outside the box" examples on Reddit like "Dan" and others to dive into.

My thinking is based on the United Nations framework and goals since it's inception by reading them myself. The so called "brotherhood of man" or "fellowship of nations" and all the rest of the socialist doctrine is in full swing with AI.

Edited August 17, 2023 by gwapofarang

[novacova]

3 hours ago, Lacessit said:

Ironic, because one of America's most cherished institutions has many socialist features and dependency - the US military.

Ironic, the above statement is completely irrational and out of context. The US military is a national security organization built to defend the US, nothing to do with socialism or this gpt/AI subject…nice try though, but just plain silly for the sake of contemptuous pointless arguments nonetheless ????

Edited August 17, 2023 by novacova

[ozimoron]

48 minutes ago, novacova said:

Ironic, the above statement is completely irrational and out of context. The US military is a national security organization built to defend the US, nothing to do with socialism or this gpt/AI subject…nice try though, but just plain silly for the sake of contemptuous pointless arguments nonetheless ????

I take it the military's human relations policies aren't socialist then? A laissez faire capital system wouldn't have them. It's exactly these policies which are under attack by the right. If the military is this pure fighting machine as you imply it wouldn't have any right wing detractors.

[Lacessit]

6 hours ago, novacova said:

Ironic, the above statement is completely irrational and out of context. The US military is a national security organization built to defend the US, nothing to do with socialism or this gpt/AI subject…nice try though, but just plain silly for the sake of contemptuous pointless arguments nonetheless ????

Attacking the messenger is a common feature of dishonest posters.

[Skipalongcassidy]

17 hours ago, FruitPudding said:

Another one:

 

Ask it to make a joke about men. It'll write a funny joke.

 

Ask it to make a joke about women. It refuses, lol. ????

 

 

Because it cannot define "woman"... data in/data out

[Eleftheros]

And there's no need for companies to worry about being flooded with job applications written by ChatGPT on behalf of illiterates seeking work, as ChatGPT-generated inquiries are easy to spot.

 

They are punctiliously grammatical, tick all the boxes of a person keen to toady to modern cultural fads like "diversity" and "inclusion" and are about as interesting and individual as rice porridge.

 

They're so good, they're awful.

[Berkshire]

20 hours ago, novacova said:

Don’t think AI can experience the essence and notion of what being an American is.

The thing is, Americans with different party affiliations can't agree on "the essence and notion of what being an American is." 

[novacova]

2 hours ago, Berkshire said:

The thing is, Americans with different party affiliations can't agree on "the essence and notion of what being an American is." 

Obviously a statement from a non American. Most Americans consider politics ancillary and respect the opinions of others, but one would never know that if they relied on the extremest views of the media for all their information. Most Americans are capitalist, self reliant, have some form of government rebellion and don’t want government interference, on all political spectrums, unlike European countries where most people are dependent on government. Only a small population in the US are extremest on both sides, just as many here that obsess on the forum day in and day out. Maybe ask AI for a rebuttal and post it here, it may bring a warm fuzzy feeling when it’s jabbed ????

Edited August 18, 2023 by novacova

[RayC]

2 hours ago, novacova said:

....unlike European countries where most people are dependent on government ????

That rather depends what you mean by dependent. Less than 1 in 5 people in the EU work in the public sector and the provision of many service utilities e.g. electricity, transport, communications, etc are, at least partly, owned by private or stock exchange listed companies. 

 

However, it's true that there is greater provision of centralised, publicly funded welfare services - such as medical care - in the EU/ UK than in the US. Imo such provision is a good thing but that's a discussion for another thread.

 

(Bold text is unintentional)

Edited August 18, 2023 by RayC
Clarification

[Berkshire]

On 8/18/2023 at 3:56 PM, novacova said:

On 8/18/2023 at 1:11 PM, Berkshire said:

The thing is, Americans with different party affiliations can't agree on "the essence and notion of what being an American is." 

Obviously a statement from a non American.

I am an American, so your premise is already wrong.  It seems you don't have a very good understanding of how most Americans think.  Which begs the question....are YOU an American?

Edited August 20, 2023 by Berkshire

[EVENKEEL]

On 8/17/2023 at 5:16 AM, ozimoron said:

lies are not true.

They are if you say them enough times.

[BangkokReady]

On 8/17/2023 at 12:24 PM, placeholder said:

You can't fathom why the response would be different. For most of the past several hundred years, being black was not a thing to be proud of. So, it's a corrective to that historical legacy. Being proud to be white has a history behind it, too. In fact, the same history that made being proud to black not so much of a thing until recently.

Or it's just programmed to give certain answers.  How can you not know this???  ????

[BangkokReady]

On 8/17/2023 at 9:15 AM, heybruce said:

If answers lean liberal, that is probably because liberals are more likely to publish their ideas and will therefore be better represented in the data fed to the program.

Or it was programmed that way by the left-wing programmers...  ????

[Bkk Brian]

You'd think those right wing techies who complain about current AI would build their own versions.

[heybruce]

2 hours ago, BangkokReady said:

Or it was programmed that way by the left-wing programmers...  ????

How much do you know about computer programming and AI?

[ozimoron]

3 hours ago, EVENKEEL said:

They are if you say them enough times.

We know you believe that already.

[sirineou]

On 8/17/2023 at 8:26 AM, KhunLA said:

No surprise there, as it seems the whole internet has be scrubbed to fit 'their' agenda.

 

Who ever controls the info ... simply controls.

I am surprised this right leaning post was not "scrubbed "from the internet. Not sure how you sneaked it in.

The scrubbers must have fallen asleep at the wheel. 

 

This article exhibits a fundamental misunderstanding of how AI works.  

 

[BangkokReady]

1 hour ago, heybruce said:

How much do you know about computer programming and AI?

I know that it's well established that Chatgpt has been programmed to avoid saying anything negative about any groups other than those of straight, White, male, Christian. ????

[BangkokReady]

2 hours ago, Bkk Brian said:

You'd think those right wing techies who complain about current AI would build their own versions.

Are there any right-wing techies?

[BangkokReady]

On 8/17/2023 at 3:23 PM, Chomper Higgot said:

There’s certainly bias.

 

Arguably the most significant technological revolution in human history and you choose to ask it questions rooted in rightwing grievances.

"It's OK for AI to be racist, as long as it's not right-wing racist."

 

You're really something else!