CHATPGT-4 in stellaris

[Stellaris enjoyer]

Is anyaone can integrate Chatgpt-4 in stellaris. For example you can communicate with other country as you want and say them what you want. Their answer will be based on the principles of the state.

[Inigma]

Avoid any callouts. Just have a stellaris gpt based AI control the game as a second player in a multiplayer game.

[Stormwind]

pete3great の投稿を引用：

I'd *love* to see Paradox have to pay out all the authors ChatGPT stole from.

Or any human authors who have read and learned from other authors!

oops...that is pretty much all of them, ever.

[Geoff]

Stormwind の投稿を引用：

pete3great の投稿を引用：

I'd *love* to see Paradox have to pay out all the authors ChatGPT stole from.

Or any human authors who have read and learned from other authors!

oops...that is pretty much all of them, ever.

Human authors do get in big trouble when they plagiarize.

[Stormwind]

corisai の投稿を引用：

Sputnik の投稿を引用：

I think you guys are far underestimating an emerging technology.

No, player's are overreacting to hype. LLM like ChatGPT are very limited in capabilities and mostly useless in current way for anything but messing around and having a fun. Plus many big failures in AI development are not known to people outside of industry.

The same situation happens with any new tech - common people overestimating it capabilities A LOT. First nuclear reactor? Fantasies about nuclear-driven robots in every house. First computer networks? Fantasies about SkyNet. And so on.

Your perspective is an important one, as it highlights the natural tendency to become enamored with the potential of new technologies, sometimes beyond their current capabilities. It's true that Large Language Models (LLMs) like ChatGPT have limitations and don't match the often sensationalized expectations. They are not sentient beings and lack understanding or consciousness, which can lead to outputs that may be irrelevant, incorrect, or contextually disconnected, depending on the input and the situation.

In the context of gaming, integrating ChatGPT could offer novel experiences, such as more interactive storytelling or dynamic NPC responses. However, expecting it to fully understand human emotions, exhibit nuanced judgment, or spontaneously generate complex strategies akin to a human player would be an overestimation. It would be more accurate to say that these models can enhance certain aspects of games where text-based interactions are pivotal, rather than revolutionizing gaming as a whole.

The historical examples you mentioned, like early reactions to nuclear technology and computer networks, show that there's often a gap between initial public expectations and the practical unfolding of a technology. In many cases, the reality is shaped not just by the technology itself, but by how it's applied, how other technologies evolve alongside it, and how societal, ethical, and economic considerations are addressed.

It's also crucial to acknowledge the 'failures' or challenges in AI development, as they often pave the way for improvements, guiding researchers and developers on the limitations they need to overcome. These instances are generally more familiar to those within the industry, but sharing these setbacks and the learning involved can provide a more balanced view of the technology for the general public.

In conclusion, while it's beneficial to remain excited about the prospects of AI and its integration in various fields like gaming, it's equally important to temper expectations with a realistic understanding of its current stage of development. This balanced view ensures that we appreciate the technology for what it is now while continuing to strive for the breakthroughs that might be achievable in the future.

[Stormwind]

Geoff の投稿を引用：

Stormwind の投稿を引用：

Or any human authors who have read and learned from other authors!

oops...that is pretty much all of them, ever.

Human authors do get in big trouble when they plagiarize.

Yes, when they plagiarize. But all authors learned from other authors, just like ChatGpt does. ChatGPT has safeguards against plagiarizing, something human authors may lack.

[Geoff]

Stormwind の投稿を引用：

Yes, when they plagiarize. But all authors learned from other authors, just like ChatGpt does. ChatGPT has safeguards against plagiarizing, something human authors may lack.

Human learning does not work the same way that automated inference does. They are very different processes. Feeding proprietary IP into an algorithm and publishing the finished results is generally outside the scopes of permissible use for protected IP. As is plagiarism. Neither humans nor engineers have carte blanche to steal other people's IP just because they have some cool use in mind for it.

[Stormwind]

Geoff の投稿を引用：

Stormwind の投稿を引用：

Yes, when they plagiarize. But all authors learned from other authors, just like ChatGpt does. ChatGPT has safeguards against plagiarizing, something human authors may lack.

Human learning does not work the same way that automated inference does. They are very different processes. Feeding proprietary IP into an algorithm and publishing the finished results is generally outside the scopes of permissible use for protected IP. As is plagiarism. Neither humans nor engineers have carte blanche to steal other people's IP just because they have some cool use in mind for it.

"Different" doesn't mean it wanders into plagiarism territory any more than a human author does. Maybe less so. You are just trying to draw the line in such a way that it will include AI.

A human may learnd from a couple soruces and be more heavily influenced by them, and borderline plagiarizing for example. AI learns from pretty much everything under the sun. SO I would argue it strays just a bit further away from plagiarism.

Try to get ChatGPT to tell you the lyrics of a song, something easily gotten from the internet. It will go a few lines then say "Thats all I can quote you, because of copyright issues".

[Geoff]

Stormwind の投稿を引用：

"Different" doesn't mean it wanders into plagiarism territory any more than a human author does. Maybe less so. You are just trying to draw the line in such a way that it will include AI.

I am not. Humans learn to speak from socialization, not from being force-fed entire canons of printed text, not from stealing transcripts of conversations recorded without consent, "Humans do it" is an industry talking point that happens to be descriptively false both about how humans learn and what generative AI does to produce the superficially human-like text it generates.

A human infant learning to speak from a human family is not at all the same as a team of corporate engineers training an algorithm on "mystery-meat" datasets.

[Exothermically Eclectic]

Geoff の投稿を引用：

I am not. Humans learn to speak from socialization, not from being force-fed entire canons of printed text, not from stealing transcripts of conversations recorded without consent, "Humans do it" is an industry talking point that happens to be descriptively false both about how humans learn and what generative AI does to produce the superficially human-like text it generates.

Humans learn to speak from listening to conversations and attempting to speak themselves, yes.
How do they learn to read and write, though?

[Geoff]

Exothermically Eclectic の投稿を引用：

Humans learn to speak from listening to conversations and attempting to speak themselves, yes.
How do they learn to read and write, though?

My mother taught me. Not all humans read and write. Teachers are not allowed to distribute protected IP to their students without paying royalties to the owner of the IP. I'll admit it happens, but every school I've attended, the administration cracks down hard on it. The current set of LLMs were not trained in the same way that human beings learn. And a lot of copyrighted material was used without consent to produce them.

[Stormwind]

Geoff の投稿を引用：

Exothermically Eclectic の投稿を引用：

Humans learn to speak from listening to conversations and attempting to speak themselves, yes.
How do they learn to read and write, though?

My mother taught me. Not all humans read and write. Teachers are not allowed to distribute protected IP to their students without paying royalties to the owner of the IP. I'll admit it happens, but every school I've attended, the administration cracks down hard on it. The current set of LLMs were not trained in the same way that human beings learn. And a lot of copyrighted material was used without consent to produce them.

lol. So only homeschooled peeps.

I have read tons of copyrighted books, I have them lining my shelf.

We all learn from each other. You have learned from copyrighted books and materials.

You are trying to make a distinction without a difference, especially when you say "your mother" taught you. So what?

This is a silly argument, frankly.

[Geoff]

Stormwind の投稿を引用：

lol. So only homeschooled peeps.

I have read tons of copyrighted books, I have them lining my shelf.

[...]

You are trying to make a distinction without a difference, especially when you say "your mother" taught you. So what?

It's a direct answer to the question. I learned from my mother. Before I enrolled in school. A perfectly mediocre American public school, fwiw. But I was one of two kids in my class who could already read by the time he started school.

And yes, it's a silly argument. Like any debate point hatched up in a corporate marketing department, it's skin-deep and brain-dead assertion. LLMs do not learn like humans do. It's a specious analogy that was introduced to water down growing resentment on the part of IP owners for the way their work was misappropriated to design these models. Along the way, it actively misrepresents the learning process of both humans and algorithms.

[Exothermically Eclectic]

Geoff の投稿を引用：

My mother taught me.

So she just beamed the knowledge directly into your head, then? No practice or anything?

Geoff の投稿を引用：

Teachers are not allowed to distribute protected IP to their students without paying royalties to the owner of the IP.

Good thing your mother never gave you any books. Having to pay IP royalties to teach her child would have been unfortunate.

[Geoff]

Exothermically Eclectic の投稿を引用：

Geoff の投稿を引用：

My mother taught me.

So she just beamed the knowledge directly into your head, then? No practice or anything?

Geoff の投稿を引用：

Teachers are not allowed to distribute protected IP to their students without paying royalties to the owner of the IP.

Good thing your mother never gave you any books. Having to pay IP royalties to teach her child would have been unfortunate.

We used a tool called "primers." I don't know if they're still standard. At some point she ran out of English ones and started using the Dutch ones from her childhood.

Reading a book is not a copyright violation. Passing off regurgitated work from another author is. If you've ever read Helen Keller's autobiography, it contains a very vivid account of the time she first plagiarized another writer and the consequences she faced for it.

I have never met a human child who learned to read by reading every book ever published. That is literally the opposite of how reading works. You learn to read, and then you read books. And even as a child, I paid for my books. And I didn't sell my writing.

They are not the same thing.

[Exothermically Eclectic]

Geoff の投稿を引用：

And even as a child, I paid for my books. And I didn't sell my writing.

Not an ambitious child, huh?
Personally, I managed to find some time for the library, but I guess it takes all sorts.