Thats what a benchmark is, and they're all gamed by everyone training models, even if they don't intend to, because the benchmarks are in the training data.
The advantage of a personal set of questions is that you might be able to keep it out of the training set, if you don't publish it anywhere, and if you make sure cloud-accessed model providers aren't logging the conversations.
The advantage of a personal set of questions is that you might be able to keep it out of the training set, if you don't publish it anywhere, and if you make sure cloud-accessed model providers aren't logging the conversations.