Notice: You have been identified as a bot, so no internal UID will be assigned to you. If you are a real person messing with your useragent, you should change it back to something normal.

Minichan

Topic: Fable 5

Anonymous A started this discussion 1 hour ago #135,621

I’m sorry but this whole narrative that Mythos is so powerful they couldn’t release it so they had to make it less powerful and release that as Fable 5 except when they released Fable 5 the US government banned it has "my girlfriend goes to another school" energy.

https://youtu.be/Y9Wz2PV404E

Do you believe Anthropic?
Poll option Votes Percentage Graph
- 0%
2 100%
- 0%

Anonymous B joined in and replied with this 59 minutes ago, 21 minutes later[^] [v] #1,441,108

It will be on Xbox and Playstation, 2027 release

Anonymous A (OP) replied with this 57 minutes ago, 2 minutes later, 23 minutes after the original post[^] [v] #1,441,109

@previous (B)
Come to think of it, it does sound like the name of a video game.

Anonymous A (OP) double-posted this 53 minutes ago, 3 minutes later, 27 minutes after the original post[^] [v] #1,441,110

Also, I just feel like… I feel like I’m nitpicking… can they come up with names that sound like what the model does? "Mythos class model" doesn’t mean anything to me.

Anonymous A (OP) triple-posted this 52 minutes ago, 1 minute later, 28 minutes after the original post[^] [v] #1,441,111

It’s like if Apple said, "the 14 inch MacBook Pro is a MacBook class laptop."

https://youtu.be/XYHLZbOag6s

Anonymous A (OP) quadruple-posted this 48 minutes ago, 3 minutes later, 32 minutes after the original post[^] [v] #1,441,112

I’m not even mad at the concept. Obviously software could find flaws in software. It’s just… this isn’t cocaine.

Anonymous B replied with this 44 minutes ago, 4 minutes later, 36 minutes after the original post[^] [v] #1,441,113

@1,441,109 (A)
They announced it 2 days before Anthropic.

You would think an AI company could generate some ideas rather than ripping off whatever showed up on their feed that day.

Anonymous A (OP) replied with this 33 minutes ago, 10 minutes later, 47 minutes after the original post[^] [v] #1,441,116

@previous (B)
Thought you were joking, it was hard to google it.

https://en.wikipedia.org/wiki/Fable_(2027_video_game)

Honestly, I don’t feel comfortable signing up for Anthropic, but I looked online and saw people saying Fable 5 made clones of Minecraft and GTA and Skyrim. I thought about it for a little bit and then I realized, where’d they get the training data to make Minecraft, GTA, and Skyrim clones? Something about that seems, possibly illegal.

Anonymous A (OP) double-posted this 32 minutes ago, 1 minute later, 48 minutes after the original post[^] [v] #1,441,117

It also feels unethical somehow, because all the code generated by AI is created by models that were trained off of stuff in GitHub without attributing any of the people who actually wrote the original code. Something about it just feels kinda gross.

Anonymous B replied with this 22 minutes ago, 9 minutes later, 58 minutes after the original post[^] [v] #1,441,120

@1,441,116 (A)
All these companies are caught pirating.

Meta (not the mexican one) was caught pirating thousands of books, magazines, and journals but justified it saying they never seeded it, so technically they didn't commit copyright infringement.

Anonymous B double-posted this 22 minutes ago, 53 seconds later, 59 minutes after the original post[^] [v] #1,441,121

@1,441,117 (A)
How would you attribute it in a situation like this?

Anonymous A (OP) replied with this 19 minutes ago, 3 minutes later, 1 hour after the original post[^] [v] #1,441,122

@previous (B)
That’s one of the issues with LLMs, you can’t really tell where the output came from. You could probably figure out where it most likely came from though just by comparing the similarity of the code to existing code on GitHub. For example, if you make a solar system simulation, it’s probably related to some GitHub repository that simulates a solar system, but it’s hard to know which one in particular because it’s probably the average of all of them.

Anonymous A (OP) double-posted this 17 minutes ago, 1 minute later, 1 hour after the original post[^] [v] #1,441,123

Even if you publish open source code though, plenty of open source licenses do require attribution even if you can use the open source code commercially in your own project. It just feels weird how strongly the US protected intellectual property until some billionaires came along and now all of the sudden copyright law doesn’t exist somehow. It really advantages the tech giants if they feel like they can use anything anybody else creates and just absorb it as their own with no credit whatsoever.

Anonymous A (OP) triple-posted this 16 minutes ago, 1 minute later, 1 hour after the original post[^] [v] #1,441,124

In general, it feels like laws only exist for the middle class and below. If I just ripped some code straight out of YouTube and made my own YouTube with code from YouTube.com I’d get my pants sued off.

Anonymous A (OP) quadruple-posted this 14 minutes ago, 1 minute later, 1 hour after the original post[^] [v] #1,441,125

Then that’s the other thing, why even use AI to write code if since you technically didn’t write the code it’s not clear whether you even can copyright AI generated code.

Anonymous B replied with this 14 minutes ago, 10 seconds later, 1 hour after the original post[^] [v] #1,441,126

@1,441,122 (A)
@1,441,123 (A)

That's what I mean, it's not something they can practically do.

Imagine asking a junior dev to attribute every influence they had. Are you remembering a stackoverflow answer? Is this you using an algorithm you read in another program?

It's not gross, because they aren't trying to hide the sources, they just can't do it.

Anonymous B double-posted this 13 minutes ago, 1 minute later, 1 hour after the original post[^] [v] #1,441,127

@1,441,125 (A)
Because people want the code to do something.

I use AI to make little scripts and apps that are useful and save me time. I can't copyright it, but that wasn't the point.

Anonymous A (OP) replied with this 12 minutes ago, 1 minute later, 1 hour after the original post[^] [v] #1,441,128

@1,441,126 (B)
Stack overflow is kinda different because on stack overflow people right responses knowing they’re giving answers to questions to people who are probably going to copy and paste it though. When you incorporate open source code into your project though, you usually actually do attribute the original authors.

For example, React JS is open source, lots of websites use it, if you include React JS in your project, you either download it or link to the CDN. Most people never see it but if you actually go to the URL and look at the code, it is attributed.

Hold up, the website is having trouble parsing the link…

(Edited 39 seconds later.)

Anonymous A (OP) double-posted this 10 minutes ago, 1 minute later, 1 hour after the original post[^] [v] #1,441,129

Not the best solution but I put it through a URL shortener I think it’s because it has an @ symbol in the original URL minichan doesn’t like that

https://shorturl.at/4iYbB

Anonymous A (OP) triple-posted this 10 minutes ago, 23 seconds later, 1 hour after the original post[^] [v] #1,441,130

At the very top of the code it says

/**
 * @license React
 * react.development.js
 *
 * Copyright (c) Facebook, Inc. and its affiliates.
 *
 * This source code is licensed under the MIT license found in the
 * LICENSE file in the root directory of this source tree.
 */

Anonymous B replied with this 10 minutes ago, 5 seconds later, 1 hour after the original post[^] [v] #1,441,131

@1,441,128 (A)
Yes, because if you are copypasting code from one source that's easy to do.

LLMs can't do it, so why would that standard be applied?

Anonymous B double-posted this 9 minutes ago, 59 seconds later, 1 hour after the original post[^] [v] #1,441,132

@1,441,130 (A)
Including a library and giving attribution and licence for it is entirely different.

There's a reason it works that way there, but not for LLMs.

Anonymous A (OP) replied with this 8 minutes ago, 19 seconds later, 1 hour after the original post[^] [v] #1,441,133

@1,441,131 (B)
Then if they’re so smart, they should invent a way for that to be done in order to comply with copyright law.

Anonymous A (OP) double-posted this 7 minutes ago, 1 minute later, 1 hour after the original post[^] [v] #1,441,134

In googles AI response on google search they figured out a way to add sources. So it can’t be impossible.

Anonymous B replied with this 53 seconds ago, 6 minutes later, 1 hour after the original post[^] [v] #1,441,135

@1,441,133 (A)
I didn't say I was so smart. You said they did something gross, I asked how you think they should do it, and you couldn't figure it out.

It's bizarre to get mad at someone for not meeting a standard, when you can't even explain how they'd meet that standard.

It's not even established in the courts that this is breaking copyright law. Fair use allows verbatim excerpts up to a certain length, and typically allows you to use heavy influence from a work if you mix it up a bit and avoid verbatim copying.

If you're going to get mad at someone or some company, you should be able to articulate what they should do.

Anonymous A (OP) replied with this 1 second ago, 52 seconds later, 1 hour after the original post[^] [v] #1,441,136

@previous (B)
I didn’t say you I said they, I’m talking about AI companies not you specifically.
:

Please familiarise yourself with the rules and markup syntax before posting.