the funimation and yunyun syndrome adaptations look very good to me.
the issue is not that an ai-generated translation "removes access" to the rom or literally prevents someone from making a better patch. the problem is that an excessive number of ai-generated translations can flood the space, lower expectations, and make unfinished or mediocre work feel "good enough". a good number of fan translations have always depended on enthusiasm, visibility, and a sense that a project is worth the effort. when a game already has an english patch available (even a rough ai-made one) many people will see the project as "done", move on, and stop talking about the need for a proper translation. that absolutely affects whether human translators feel motivated to pick it up later.
this is especially true because fan translation is unpaid labour. translators are not usually choosing projects because there is a financial incentive. they choose them because there is passion, community demand, and the feeling that their work would matter. if the first thing people find is an ai patch that technically makes the game playable, the urgency around a proper translation often disappears, even if the quality is nowhere near what a human team could produce.
and yes, people can still make their own translations. but in practice, i have almost never seen ai translations actually get replaced by proper human-made translations afterward. more often, the ai version becomes the de facto translated version, and the game never receives the care it deserved. the only case i can think of where that happened was the 3ds version of ex troppers, and even then, the later human translation did not seem to spread nearly as much as the original ai-translated version. that is exactly the problem: once an ai patch becomes the first widely available option, it often becomes the default version in people's minds, even if a better human-made translation appears later.
also, i don't think comparing this to the old days of babelfish-style fan translations really works anymore. back then, the scene was smaller, tools were worse, and many people were just doing whatever they could with very limited resources. today, expectations should be higher. we have better documentation, better hacking tools, more experienced communities, and many examples of excellent fan translations. so the question should not be "is this better than nothing?" forever. at some point, quality has to matter.
that said, even though i am mostly against the use of ai in this context, i did try creating a machine-translated version of digimon story xros wars for the ds. it translated all of the game’s dialogue in a very... ai way. some lines became simply incomprehensible or completely lost their meaning. i would never release that version publicly; i made it for myself, precisely because i do not think the excessive spread of ai-made translations is a good thing. in the end, i gave up on it for the same reasons mentioned above. what can be done, and what i fully support, is using the necessary ai tools to assist translators and rom hackers in carrying out their projects. there is a big difference between ai being used as a tool within a careful human-led project and ai-generated patches being mass-produced and treated as acceptable substitutes for proper translation work.
