4 ms·
You’ll need a whole new class of training data. A camera strapped to someone’s head as they go about their day, scaled across continents, age groups and occupat
by leobg 2y ago
You’ll need a whole new class of training data. A camera strapped to someone’s head as they go about their day, scaled across continents, age groups and occupations. You can’t rely on YouTube, since that’s all edited. Not at all how a toddler would see the world.
- ehnto 2y agoI partially agree, but many humans do infact learn new languages from youtube, TV shows etc. There is also practically infinite minutes of long form conversations on youtube at least in English and Japanese. Not edited just 1+hour long videos of uncut talking. Funnily enough though, the comprehensible input movement for language learning has already started generating the type of content you're suggesting. Long first person view walks, following day to day activities and usually with a monologue about what they are seeing inbetween.
- vasco 2y agoHave you heard of live streaming and lifestreaming? There's countless hours of this exact type of footage available on twitch. I would bet Google is also already recording when doing mapping, so they should have videos going through every street in the world as well.
- nosrepa 2y agoA lot of people don't know that twitch started as justin.tv and was specifically a life streaming platform.
- realusername 2y agoI've been learning a language from Tiktok videos for more than two years and that totally works, the results are there. This format is even more edited than YouTube
- leobg 2y agoYou also got trained by ~5k hours of unedited real-time first person footage before you started to speak.
- realusername 2y agoNo I wasn't, this is my only exposure to this language, I don't live there.