I have been looking at what data it produces for a few days now, and it’s using 2015-2017 data for it. I have come to this conclusion because of me attempting to get a group 16, and it spits out older youtubers (some who have been very much canceled). If you can attempt to update the data for 2022, that be nicer for it to gain fresher sets and lingo. Thank you.
It isn’t that easy as editing a JSON file to update data on recent events.
Since this is a huge neural network, it needs to be retrained, which takes a lot of time and money.
They likely won’t retrain it just to gather new events. In fact, they even stated it hee that the data ends at about 2021.
What you could do is include relevant data inside the prompt and it’ll use it. In your case include summary of some YouTubers and then ask it something about them.
If you have a lot of them and don’t want to use 3K or whatever amount of tokens with every request, consider abstracting it into multiple prompts (requires programming) or simply fine-tune the model to “make it learn” about new YouTubers.