This was always going to happen at some point. Current models are just brute force(more data, more compute), it shouldn't be shocking that someone developed a more elegant method. It seems pretty dumb to think that the current path was going to continue to be the correct one.
|