Re: In the spirit of the story...
"fun is in reading YouTube's machine generated subtitles".
And there was me thinking I was the only one who noticed.
Part of my aversion to 'AI' or 'ML' is caused by the output of whatever generates these subtitles. You sort of expect errors in translation, and the 'Audio to text' task is hard in the first place, but the audio--> text error rate is far too high for my liking. It's only the odd word missing or mis-translated but this is usually more than enough to skew the text into something a long way from the audio. I guess the 'English Audio to {Other language}' task is harder and more prone to error, especially as us flesh-sacks can't do that properly either!
It's one thing to argue about translation with a human.....but how do you argue with a machine about it's translation? Personally, I can't wait for someone in the legal community to submit/try to use some machine-translated text and expect it to remain unchallenged.
Biting the hand that feeds IT
