I think a good discussion on this will be quality vs quantity. If we're talking about self driving data, I don't think anyone is beating Tesla there. All the info and data that they got from their cars will give them an advantage over their competitors. But when we talk about LLM and Generative AI, Tesla's data provides a very specific and limited type of data. Social Media has all sorts of data for AI training. Just looking at YT, there are cooking tutorials, art tutorials, video games, math/science/english channels, etc. Twitter's data will definitely help the missing data from Tesla.
Sort: Trending
Remember, computers do not see cooking or driving videos. Machine language is binary. Everything it is fed is converted to binary language.
If that is the argument, then what is the point of training them on all sorts of data? Why don't they just feed them random binary? The cooking and driving videos are still different data when converted to binary. Variety is still important.