pokedude900 posted...
You keep making sweeping statements like that while disregarding any actual arguments. How about responding to post #25?
you dont think the model could be trained to parse things like a simple swap and replace? lol
or that certain phrases could be flagged ahead of time? (ie: movie quotes). human translators get shit like that wrong without notes too.
puns, etc. seem like a steep challenge, but again, human localization rarely does a good job either (because there's rarely any way
to
do a good job without TL/N)
and you're basing your assumptions off of current AI capability. i wholeheartedly believe in the next decade or so we'll have AI translation that's mostly indistinguishable from hand done, and i think it's absolutely foolish to say it will "NEVER" happen