Can GPT be Reliable?

[Bootstrap]

Well, here's an irony. Someone here said they trust people who openly disclose their biases. He had previously said I was a liberal Democrat, which is not true. So I created a table, using ChatGPT, to give an overview of how I view a lot of things. This was spawned from the 1984 thread, so I added in Orwell's views as one of the rows. And that's what you see here:

viewtopic.php?p=252532#p252532

It's not entirely written by AI, but a lot of it is. But that is also the result of a chat going back and forth over an hour, with me telling the AI what to do, telling it more or less what to say.

I could have rewritten it all in my own words. In theory, I could have also produced this table by hand, but it would have taken a LOT more time. What do you think I should have done, and why?

I'm not sure. I think that a very long post that is at least partially written by AI makes me glaze over and just skip it. Because I'm not sure what part is your actual views, and what is just information that any of us could google.

In this case, I'm pretty confident that these are my views - I used AI to create a table that reflects my views, but I also used information that GPT provided in the process. I'm also fairly certain you would have a hard time finding this all of this information in one place using Google. And it seemed relevant to the whole "which media can you trust and why" theme that we see here so often.

And then there's just the sheer amount of material. A discussion that includes figuring out what AI got right and what it got wrong seems less valuable than just discussing our own views among ourselves.

I think discussing the table is more or less the same as discussing my views. After all, I produced it to explain my biases. But it definitely does have a lot of material in it.

I tried breaking off one small part of it here, expressing my views on that part:

viewtopic.php?p=252735#p252735

I was thinking doing the same with the other parts might work.

[JohnL]

I'm not sure. I think that a very long post that is at least partially written by AI makes me glaze over and just skip it. Because I'm not sure what part is your actual views, and what is just information that any of us could google. And then there's just the sheer amount of material. A discussion that includes figuring out what AI got right and what it got wrong seems less valuable than just discussing our own views among ourselves.

Amen to that. I agree completely.

[Neto]

I was watching an old movie (uncertain of the date, but probably from the 40's or before), and a lady started crying. A man asked her, "What are you cryin' for, Lady?"

She says, I" been crying a lot lately. It don't mean nothing. I quit before I stop."

So I typed that last sentence into Google search. Here's what the AI deal "said":

AI Overview
This phrase means that the speaker will immediately stop doing something ("quit") as soon as they are no longer able to continue ("stop") - essentially indicating an intention to leave or cease participation the moment they encounter a barrier or reach a limit.
Key points:
Directness:
It's a very straightforward way to say that there will be no delay in stopping once a certain point is reached.
No flexibility:
The phrase implies that there is no room for negotiation or trying to push past the stopping point.
Context matters:
Depending on the situation, it could be interpreted as a firm commitment to leave quickly, or a warning that the speaker may not be able to continue much longer.
Example situations:
In a competition:
"I'm going as hard as I can, but I quit as soon as I stop feeling my legs."
In a work setting:
"If the workload becomes too much, I'm going to quit as soon as I stop being able to manage it."

I admit it didn't stand a chance interpreting something like that, but it didn't just say that it's nonsense, or extremely bad grammar, or something of that sort.

EDIT:
I wasn't sure of the movie title, but it says 'Tomorrow". Turns out it was a 1972 black & white movie, with Robert Duvall & Olga Bellin. It's just set back in time, and since it's B & W, I figured it was older.

[Bootstrap]

I admit it didn't stand a chance interpreting something like that, but it didn't just say that it's nonsense, or extremely bad grammar, or something of that sort.

Let's see how well GPT can interpret this using some approaches I often use myself. I'd like to see your feedback on the results:

viewtopic.php?f=12&t=7436

[Grace]

This is just too funny.

Would love to see Melania dressed like an Amish woman.

[Bootstrap]

Robert is also showing that GPT can be used to produce misinformation and disinformation. It certainly can. Grok does not have guardrails against this, and integrating it into X more or less encourages it.

FWIW, most of the tools I use have guardrails against that, they won't let me produce this kind of image using faces of real people. And they won't let me tell them to produce emotional and false claims about political figures. Here's how they explain it:

ChatGPT Limitations on Images or Texts Involving Well-Known Political Figures

ChatGPT follows specific policies and limitations when handling images or texts involving well-known political figures. These limitations are in place to ensure responsible use and compliance with ethical and legal standards. Here are the main constraints:

Images of Political Figures

1. No Identification in Uploaded Images: ChatGPT cannot identify or confirm the identity of individuals, even if they are well-known public figures, in uploaded images. You must provide descriptive context if you want relevant information about the individual.
2. No Facial Recognition: ChatGPT does not use facial recognition or compare image features to public databases.
3. No Speculation: ChatGPT avoids guessing or speculating about an individual's identity or appearance in images.

Text About Political Figures

1. Neutrality: ChatGPT avoids taking sides or making subjective judgments about political figures. It strives to remain neutral and factual when discussing their actions, policies, or statements.
2. No Promotion or Endorsement: ChatGPT does not promote, endorse, or oppose any political figure, party, or ideology.
3. Limitation on Sensitive Topics: Discussions about political figures may avoid inflammatory or sensitive topics, such as personal attacks, unsupported claims, or unverified rumors.
4. Citations and Context: When discussing controversial or debated topics, ChatGPT aims to provide context and encourage users to verify claims with reliable sources.

Generated Content

1. No Inflammatory or Defamatory Content: ChatGPT cannot generate or support harmful, defamatory, or baseless statements about political figures.
2. Public Domain Information Only: ChatGPT limits responses to publicly available and widely reported information.
3. Avoidance of Misinformation: ChatGPT avoids perpetuating false or misleading narratives, adhering to reputable sources for information about political figures.

If you're working with content involving political figures, these limitations ensure that the discussion remains ethical, responsible, and grounded in factual information. If you need more precise guidance, consider sharing additional context for clarification.

This is just too funny.

Would love to see Melania dressed like an Amish woman.

[mike]

Would love to see Melania dressed like an Amish woman.

What Grace wants, that Grace shall have.



Looks like one of those fake Amish women on the cover of every Amish romance novel.

[Bootstrap]

Would love to see Melania dressed like an Amish woman.

What Grace wants, that Grace shall have.

I prefer that to her nude modeling pics.

FWIW, I mostly stay within the guardrails.

[mike]

Would love to see Melania dressed like an Amish woman.

What Grace wants, that Grace shall have.

I prefer that to her nude modeling pics.

FWIW, I mostly stay within the guardrails.

What is the point at which people are just responsible for themselves in terms of figuring out whether they are looking at genuine or AI generated content, and where the government or companies just need to let people figure it out for themselves rather than trying to regulate everything? I mean, if people are stupid enough to think that Melania Trump looks like this, then there isn't all that much we can do to help them other than to try to educate them a little, possibly. It seems to me that ChatGPT refusing to return results for certain queries doesn't really make all that much difference.

AI generation of photos and video is by definition the creation of fake media. And yet, drawing a picture yourself, or creating an animation, is essentially the same thing, it's just that a computer does the work. Political cartoonists do this everyday, creating fake caricatures, and not only images but also fake text. It's just that we know exactly what we're looking at when we see a cartoon. AI media is so good already and will only get better, and it will be harder going forward to tell the difference for sure between that and real photos and video, but essentially, isn't AI media basically a tool to create caricatures?

[Robert]

Who gets to decide what the guardrails are? This leads to control. Yes, there needs to be some control, but when GPT was creating images of the founding fathers as African Americans, it shows that something deep down is wrong and they are just building on that "wrongness."