The gap between what Google should be able to do and what they regularly end up doing has got to be the biggest in the game. It's a bit odd given that they're Google, yet OpenAI routinely beats them at stuff like this. I suppose because it's free and available to everyone the model has to suck?
It's becoming clear that (for a while at least) there will be a massive difference between those who pay for the good models and those who don't (or can't).
|