It's actually even worse than that: the current trend of AI is transformer-based deep learning models that use self-attention mechanisms to generate token proba

It's actually even worse than that: the current trend of AI is transformer-based deep learning models that use self-attention mechanisms to generate token probabilities, predicting sequences based on training data. If only it was something which we could ontologically map onto existing categories like servants or liars...