安装 Steam
登录
|
语言
繁體中文（繁体中文）
日本語（日语）
한국어（韩语）
ไทย（泰语）
български（保加利亚语）
Čeština（捷克语）
Dansk（丹麦语）
Deutsch（德语）
English（英语）
Español-España（西班牙语 - 西班牙）
Español - Latinoamérica（西班牙语 - 拉丁美洲）
Ελληνικά（希腊语）
Français（法语）
Italiano（意大利语）
Bahasa Indonesia（印度尼西亚语）
Magyar（匈牙利语）
Nederlands（荷兰语）
Norsk（挪威语）
Polski（波兰语）
Português（葡萄牙语 - 葡萄牙）
Português-Brasil（葡萄牙语 - 巴西）
Română（罗马尼亚语）
Русский（俄语）
Suomi（芬兰语）
Svenska（瑞典语）
Türkçe（土耳其语）
Tiếng Việt（越南语）
Українська（乌克兰语）
报告翻译问题
讨论规则及指引
2 
举报此帖
Not all developers can afford to pay other people to do the work. A one-man team who cannot afford an artist, would naturally see generative art as a good compromise. Which, if that model has been ethically trained, isn't such a bad thing. A human artist would be preferred though. Same with music assets.
When it comes to writing, there's no harm in having an LLM proof your work, or using it to bounce ideas off of, as long as you're aware of the errors the LLM can make and of the constraints it is running under. To many safety rails definitely stifle their abilities with creative writing.
But I do agree with your sentiment when it comes to those who absolutely can afford to hire people to do a better job. I'm looking at you, Activision.
Your caveat is, at this time, not possible. If it was I might have agreed, but I have yet to read of any sufficiently good and ethical GAI stuff. For anime styled games there are at least somewhat ethical alternatives.
As far as writing goes I just think back to that debacle where an author drew hype for having a latina heroine until he announced he'd used AI a bit too much (or something) and her name was Latina. I wonder what happened with that one LN.
I do somewhat agree with having it proof your work, but everything else is a bit excessive IMO. I want to read good ♥♥♥♥, not AI ♥♥♥♥.
Now what I do agree with is that the big players shouldn't utilise the plagiarism machine whatsoever. There are good parts to AI, don't get me wrong, but GAI is just no.
But one way to you could use it is say model a tree. And then have AI make thousand variations of that tree. Not really a benefit for a person to make a thousand variations by hand. And it's not like it's so different from procedural generation schemes that have been in use for decades.
You can use AI to reduce some drudgery or boilerplate sort of work.
I think any time there's new tools and technology people have concerns about how it will impact them.
In software development (non-gaming) some developers worry about AI stealing their jobs, and some C suite types probably salivate over the idea. But.. when a non-programmer can provide decent requirements and specs and user stories to AI, well I have it marked on my calendar, the 5th of Never.
The idea versus reality isn't all roses, and the devil is in the details. But I have used AI a little bit to spit out boring boilerplate code and that's pretty nice. I can focus on the specific features and functions specific to the program, rather spending time remembering the details of writing a string parsing function to detect diacritics or whatever problem that has been solved a hundred thousand times in every language.
Point is there's lots of valid and appropriate ways to use AI that don't involve theft or copyright violations or trying to drive artists to extinction.
I mean I knew a guy in elementary school whose first name was Latino. His parents were a black and white interracial couple, so someone having the name Latina isn't exactly something limited to AI doing it.
https://x.com/fspls/status/1824374115962327233
In retrospect it was a pretty funny situation, but I wonder how much he used AI in his "research".
There is a lot of game features that use it the normal user will know nothing of and if it's just over the recent "ai art" issue that falls on the user to deal with but more and more games will use whatever methods that make creating games easier and cheaper and it's not going anywhere.
At some point companies will be required to disclose they used humans.
Before we got on the "AI AI AI" train, procedural generation had been going on for awhile. We've had auto-generated worlds before, Age of Empires' Random Map, Civ, Elder Scrolls II: Daggerfall, but it wasn't really until Minecraft that generating worlds actually showed how cool it could be, and that tech has just gotten better.
The issue starts to come in when AI gets used to generate other things. Age of Mythology was grilled during the beta when the new god portraits used AI... and you could tell. Like, this isn't some small studio either, this is a 25+ year running series that's now being run by Microsoft, and your idea of Bast is a fuzzy house cat's head sitting on a dainty human body? They ended up redoing the portraits with actual artists, and ho crap, she looks GOOD now.
AI being able to talk to you or create complex drawings quickly is certainly interesting new technology, but the issue is always when it's used out of laziness, as you start to get strange mistakes, artifacts, and 'uncanny valley' moments.
Also, AI has one pretty big glaring weakness right now: it needs to analyze human-made content in order to generate its own content. One AI can't learn to draw better from another AI, it just ends up copying the same mistakes and applying the new mistake to its logic, so it needs a regular supply of human-made content in order to keep going.
it's literally just a algorithm. that steals other peoples work and smashes it together to make an image.
One Japanese person being able to write, draw, etc., doesn't mean the vast majority of people of any nationality can. Most can't. In fact, Japanese, it could be argued, have an advantage in that area due to the wide acceptance of anime and manga, and the prevalence for being encouraged to draw in that style -- however poorly they manage. Higurashi's author's art may look bad, but most people can't even manage THAT much. Also, schools and the like in other countries can be outright hostile to students wanting to draw in such a style, so they lack the chances and encouragement to improve.
As for ethically trained models, they are out there if you look. Plus, as more countries start to consider how to treat unethically trained models -- and the resulting products, we'll likely see new avenues for ethically trained models to be sold. Don't be surprised if, in a few years time, such models are being sold even here. They will, ultimately, be little different to the vocaloid software and voice packs that include Hatsune Miku. Except where Miku is for music, these models will aid with art.
So, in short, I think you're being a bit restrictive when it comes to people who do not have required skills, cannot obtain them -- no matter how much they try -- and cannot afford to hire those more blessed in that area. For these people, generative art, music, etc., is a tool they shouldn't be punished for if they are being ethical with it.