AI-generated content: Legality, ethics, and the nature of art

" Are the engineers who built Google Search liars because Google Search sometimes displays misleading results?"

Depends on the details. Did they know this would happen? Did they take steps to discourage it? Does the search algorithm itself lead to misleading results? Are the search results biased in some way, that Google engineers should be aware of?

The Google algorithm is well known to be biased in various ways (ironically, you can determine this by Googling it). So I would say "Be careful using Google, because the results may be lying to you" is not wrong.

"Now, when there's overlap between the people engineering it and pushing for its usage, then they're lying, but the blame for knowingly leaving misinformation with some buried disclaimers should be going on the executives and such who want to use the products for everything possible, or push it as part of a search engine."

Ok, I can agree with all that, but it still behooves us to teach the public to approach these results with a healthy degree of skepticism.

Edited by DeMarquis on Jun 4th 2023 at 3:51:46 PM

The people designing a search engine are intending to make a search engine, and as such fairly possess blame for it being incorrect (intentionally or otherwise). It's a measure of the intended success of their product whether or not it gives people the information they're looking for.

An LLM is still a chatbot, it's just one with an absolutely enormous data corpus. While this means it can answer a lot of questions properly, it's not the thing it's designed from the ground up to do. Replacing ChatGPT with 'a general purpose chatbot' in most of these questions makes it pretty clear why it's odd to pin it on developers.

Edited by RainehDaze on Jun 4th 2023 at 9:05:34 AM

Avatar Source

When Google Search goes beyond just showing links to websites based on your search terms - which it does now; it shows Q&A snippets related to your search - if it’s choosing those from websites whose information is wrong, then yes, you are being lied to. You are being told “look, here is the right answer to your question”, when that is not the case.

Browsers are now using Chat GPT as an AI for the purposes of proving infomation and answering users’ questions. That is not its purpose and not something it is capable of competently doing, so building it into the browser and acting like that’s what is IS for is flagrantly deceptive.

Edited by Galadriel on Jun 4th 2023 at 1:55:32 AM

Yes, and do we agree that this deception isn't the fault of the developers who made a really good chatbot (at least, not in their capacity as developers)?

Avatar Source

Provided the developers didn't know the ultimate purpose, then yes, they would be blameless. But that comes down to matters of intent. Many of the early pioneers in the history of statistics were trying to design a way to prove that certain races or ethnic groups were superior to others, so they bear blame for their bad intent, even though the tools they provided us with work perfectly well for many other purposes. Then again, most statisticians were perfectly honest in their desire to simply improve humanity's ability to process data.

So I'm afraid the devil is in the details.

I don't think I've ever seen any indication that the purpose of people studying Large Language Models has been 'spread bizarre misinformation', so...

Edited by RainehDaze on Jun 4th 2023 at 12:03:10 PM

Avatar Source

The programmers almost certainly thought that no one could possibly be dumb enough to take them as an authority.

Not Three Laws compliant.

I can guarantee you bings engineers were probaly screaming into the aether as they were told to add a chatbot to it.

That is 100% on marketing.

A chatbot adds nothing to a search engine from a functional level and its addition goes agianst the best practices of KISS

Does it make it more accurate? No it actualy makes it less so.

Does it make it easier to use? Not really, search engines tend to handle natural language questions just fine in the overwhelming majority of cases.

So why add the needless complexity.

Because managment and marketing can see google doing better then them and want the next big gimic and wont listen to you.

Edited by Imca on Jun 4th 2023 at 6:33:38 AM

And the last time anyone really cared about Bing was when it was a shortlived meme after Microsoft's disastrous Kinect presentation. The one where the poor model had to put on a weird accent when ordering it around and it was like "Xbawks BEEEEEEEENG".

Not Three Laws compliant.

Citation needed

EDIT: As a programmer, you are supposed to assume that the user is an idiot. Mind you, it is not about looking down in users in general. The problem is that you can never know what people may do with your program as a result of stupidity or genuine mistake. You have to prepare for everything, and assuming the user is stupid helps with that.

Edited by Risa123 on Jun 5th 2023 at 10:37:57 AM

…really? I really need to cite the idea that the programmers for a chatbot didn’t think it would be used like it’s a source of reliable knowledge? You’re serious?

I think my citation is that it doesn’t work as a source of reliable knowledge because that’s not why it exists. If the programmers thought it was supposed to be a source of reliable knowledge, they would have programmed it to be one.

And assuming the user is an idiot has limits. Programmers are notorious for overestimating users.

Edited by Zendervai on Jun 5th 2023 at 4:39:23 AM

Not Three Laws compliant.

That was not meant to be literal. I meant that you have to back up your claim.

And another claim out of nowhere. I mean, you could argue that as a programmer, I'm biased. That said I do not think that programmers are perfect, but seriously ? That is easy for you to say.

Edited by Risa123 on Jun 5th 2023 at 10:43:57 AM

Also as a programmer, I don't think I'd ever expect that I should be calibrating a chatbot to be treated as an oracle with all the answers in the world.

Avatar Source

From the point of view of the public, it's less important exactly who lied than that someone did (and will likely continue to do so in the future).

Ah, but from the point of view of a thread that's 2/3 pedantic arguments?

Avatar Source

Point made. : )

Edited by DeMarquis on Jun 5th 2023 at 8:44:25 AM

A Twitter account has translated a Japanese Copyright law regarding AI. AI can be used for research and education without infringing on Copyright, up to a certain point.

Copyright holders are given free reign to take legal action on those who break the law and sell reproductions made through AI

Here’s the Japanese link which also talks about said Law regarding Ai and Copyright. Site is PC.Watch

Edited by OmegaRadiance on Jun 6th 2023 at 7:16:01 AM

Every accusation by the GOP is ALWAYS a confession.

This is honestly very surprising given what I've seen from members of the Diet who have been speaking up in favor of AI, including Akamatsu Ken and Kawasaki Hideto, both who have been vocally in support (Kawasaki in particular has his own line of AI NFTs).

I'm wondering if this is some too-good-to-be-true phrasing that's going to be interpreted as loosely as possible, or otherwise place an unreasonable burden of proof on artists.

It's been fun.

Kinda seems like the law is just making it clear that normal standards of "don't reproduce someone else's work" applies to AI-generated stuff too. The law specifically targets AI works that are "uploaded, published, or sold as reproductions" and those that are "almost identical (derivative) or CLEARLY dependent on existing copyrighted works to produce a 'new' work." I think "clearly" probably applies to people who prompt generators to make a work in the style of X artist, but probably not to generic generations.

There is no war in Ba Sing Se.

That is correct, because the previous version of the law was ambiguous about that sadly.

This is mostly in Filipino, but it's quite insightful and balanced it ways in using it wisely. My brother made it more insightful when he compares AI with guns. It's not that AU or guns that are inherently evil, it's the people who use it.

ᜇᜎᜈ᜔ᜇᜈ᜔|I DO COMMISSIONS|ᜇᜎᜈ᜔ᜇᜈ᜔

OK, mind blown. Someone has combined three different AI apps into Skyrim, the game. You can now have an extended, unique conversation with any NPC in the game. And yes, one of the AI apps is trained on the conversation trees already in Skyrim, so the NPC's respond in character. And these are not short conversations.

ESO (the guy who made the video) has some interesting things to say about the future of AI in video games.

Edited by DeMarquis on Jun 16th 2023 at 2:40:14 PM

I dread ever having to speak to play video games.

Avatar Source

I have to be honest, I can only really see that working in a game with a ton of basically irrelevant npcs. Like yeah, it can be helpful, but it's not a great system for games that aren't open world or that are designed to have the npcs give you specific information.

Not Three Laws compliant.

This is a use of AI that catches my interest. The Videogame AI (normal program with no connection to AI that is subject of this thread) has a reputation for stupidity. So developers often have to resort to let it cheat to ensure it can compete with human player. It is not just about more interesting dialogue, but perhaps we can also use AI to make games more challenging.

May I ask why ?

Good point.

Edited by Risa123 on Jun 16th 2023 at 8:54:45 PM