Brilliant bot.

I didn't understand what you meant about mods. What are you getting at?

I mean that the popularity of some characters increases over time, while the popularity of others decreases. Accordingly, the program tries to suggest more popular characters, for which the algorithm is better developed (due to the large number of iterations), and does not suggest less popular ones (assuming that they will still not be guessed, and if they are guessed, the algorithm will most likely not work very well and will lead to some non-standard situation that is not in the database => a dead end). Therefore, the probability that Akinator will guess a "not trendy" character is not very high, although it most likely still exists, but it is rarely realized. I assume that the algorithm there is probabilistic, because it somehow needs to reduce the number of questions leading to the answer, otherwise it will take too long.

Yeah, Akinator consistently suggested other dragons, and even twice from Skyrim. And he asked several times "Is this a character from Skyrim/TES?", to which I answered "no". He also suggested characters from Final Fantasy, although I also answered "no" to that.

It's a feature of the training process. Apparently, he is very often asked about dragons from Skyrim, even without knowing that they are from Skyrim.