The translation that shows up when clicking on a phrase is different than the one that the Explain tab gives. In this example you can see the clicked translation is just wrong, even though ChatGPT understands the correct translation from the context in the Explain tab.
My understanding of context aware translation is to see the definition provided by the Explain tab directly when clicking on text, without having to actually open the Explain tab. Is this not the intended behavior?
I would’ve expected to see either “the divine lineage” or “the families of gods” over gudeslektene, instead of “the god”.
It’s worth just double checking that you have “Context-Aware” selected from: reader page → sidebar → AA → Translation Behavior.
But even so, this kind of thing isn’t completely unexpected. The AI prompt used to generate the context-aware translations is not the same as the one used to generate the explanation. The context-aware translation is intended to be a single translation of the word you clicked on. The explanation has more room to provide multiple, perhaps longer, translations and explanations. And the somewhat unpredictable nature of the AI used means that the translation(s) provided within the explanations are not guaranteed to contain the exact translation provided by the context-aware translation. Most of the time it probably will but it’s not guaranteed. In the case where you prefer a translation provided within the explanation I suggest clicking the button, e.g. “+ the families of gods” so that that translation will be added to the generated flashcard.
