> Also, since I assume MDN has a finite number of things that need explaining, why not let users mark an AI explanation as wrong, and submit their own corrected

> Also, since I assume MDN has a finite number of things that need explaining, why not let users mark an AI explanation as wrong, and submit their own corrected versions? I mean if you implement it well you don't really need the model. If anything it would be more effective if it was always just slightly wrong.