The speaker says deep underlying model properties transfer across generations because each generation is trained on data from the previous generation.

“And so there's some like deep underlying properties of the model that are being sort of transferred between model generations, because basically you train your AI on data from the prior generation and keep going.”