The prevailing wisdom is to try as much as possible to avoid using pre-generated lists or just copying someone else’s list. The reason makes good sense, if you come up with a list yourself, you are taking advantage of pre-existing neural connections.
Example, If you see something like 913, and sound it B-T-M, the first thought you have may be BaTMan. You see those numbers and your instinct is to sound it out that way. Your chatGPT list, or some random person you’ve copied may interpret 913 as P-D-M and have it as Padme, from Star Wars.
If you try to re-wire your pre-existing preference for 913-BTM-BaTMan so that it matches a word or image you didn’t come up with, it will take more time to transition to that other image and the neural instinctual connection won’t be as strong.
It’s super tempting to take the shortcut and just fill in a few gaps yourself, but I’d encourage you to take the opposite approach, only use it as a last resort when you’ve exhausted your brain for a few days/weeks trying fill in a single word here or there. Yes, it takes longer, but it’ll pay off when you get to the association phase. Even then, when you practice you may see a card pair, sound out the consonant sequence and a completely different word/image springs to mind, which you just keep coming back to every time. Don’t be afraid of changing words of your brain keeps offering a more consistent one. This is especially true if you have personal connections with people or unique objects that AI wouldn’t have access too, say your favorite uNCLe for 275… Your personal image would certainly be more vivid and connected to you than a “NiCKLe” that might be generated from someone else’s list.
The only challenge with revising is making sure your images are unique between each other. I realized I had several basketball players in my list and ended up changing DuRaNt to TRoN to make it stand out more. That was after a week or two of practice seeing 142 as DuRaNt. It was annoying to re-associate, but worth it for the variety and vividness of the eventual image.