Despite still being in its infancy, Machine Learning models of AI have demonstrated that it is humanity’s next big innovation. The use of AI for translation and asset development has already greatly impacted something as niche as VN development.
Recently, there’s been much debate about the use of AI in the creation of art, music, even voice acting, and nearly any kind of creative medium you can think of. A handful of VN developers have either used AI itself to make assets, or to assist in doing so.
This thinkpiece breaks down the pros and cons of AI, to highlight key points for you to decide whether AI has a future in creative endeavors. If it poses a risk to human creativity and the rights of artists, or somewhere in between. If you’re not convinced that the use of AI in the creation of VNs (and all artistic mediums, really) is an issue worth addressing, or don’t understand why its such a big point of contention, this piece should hopefully shed some light on the AI phenomenon.
Our community’s opinions
A few weeks ago, I made a thread on Fuwaforums asking for opinions on the use of AI in creative projects, particularly VNs. The responses to the thread were fantastic! Many commenters had deep opinions on the topic and presented great points. I encourage all who are interested in this topic to continue the discussion there.
“I’ve seen the AI-generated games that are popping up on DLsite, and I even played a few. My first thought after doing so was… boredom. They look beautiful and technically hit the fetish points of those trying to play them, but they somehow manage to be completely bland and soulless nonetheless. “-Clephas, Fuwaforum moderator
Some pros include:
- The use of AI upscaling is seen as a positive, and is something few take issue with. In these cases, it wouldn’t be possible to have a higher quality asset without remaking it from scratch.
- Creators can use AI for placeholder assets.
- Developers with tight budgets can use AI and still have decent art for their projects.
- Quick and easy assets for game jams, or smaller, passion projects.
Some cons include:
- AI is terrible at writing.
- Music, sound, and voice acting generated by AI is also awful.
- The ethical issues of AI being trained off artists work without their consent.
- Developers using AI instead of hiring humans for jobs is a con in itself, and worthy of another discussion.
Based on the responses in the thread, we can see some key points to help gauge where AI is now
- Aside from the occasional bad hands/finger/teeth AI artwork is decent.
- Music, SFX, voice, etc content made by AI is very poor quality.
- Coding can be competent (but low-end).
I also asked users to fill out a poll to get a more visual indication of the opinions surrounding AI in VN development.
The poll’s results show that a plurality of voters were neutral, or against the use of AI in VN development.
AI and MTL in VN translation
Another important potential use for AI in VN development is translation. Like with AI art, or image upscaling, AI translation is something that has already started to gain traction.
It’s important to make a distinction between AI translation and MTL. Machine translation (MTL) is like an algorithm of sorts, it analyzes patterns from both languages, and can decide how to translate a sentence or phrase based on probability.
Translation that uses AI, on the other hand, goes many steps further and can use large amounts of data to help translate slang, or things that may need more context. Something like DeepL, for example, uses both MTL and AI. That being said, both types of translation have the same impact, so I’ll mention them interchangeably.
The biggest issue with AI/MTL is context, which is especially important in a context-based language like Japanese. Something like Google Translate or DeepL may be fine as over-the-counter tools for translating web pages or Social media posts in another language, but human intervention is needed for works that are longer, or require more attention to detail such as fiction, legal documents, etc.
MTL is not always as precise as human translations.
It’s entirely possible for developers to begin translation using AI/MTL, then have a person look over it afterwards to make corrections, or translate a word or phrase better. This process may even become the standard in the future, but at that point, it may be more efficient to have a human translator in the first place.
I think we should be cautious of the use of AI. Aside from placeholders, upscaling, or the low-end side of art or coding, it seems likely to have a negative effect on creative staffs and could decrease the overall quality of a product based on the current known limitations of AI, such as with translation. One could also make a more philosophical argument against it, perhaps articulating that life could become too easy if AI advanced to a certain point. However these implications go far beyond the use of AI in the development of Visual Novels.
Either way, given the fact there’s already major controversy surrounding the use of AI, its unlikely the progress and usage of AI will get to the disastrous end point people fear. Optimistically, we may find its place in creative works, wherever that may be. Even if AI makes VNdev smoother, it still takes a competent team to make the final product worth reading.
If you enjoy our work, please follow us on Twitter for more visual novel news and join the community in our forums and discord server: we’re pretty active there!
In addition, you can also support us on Patreon or by buying us a Ko-fi.