3 ms·
That might be smarter than we initially give it credit. By leaving the “safer” (read: harder to get wrong) things to their own models and then the more “creativ
by latexr 2y ago
That might be smarter than we initially give it credit. By leaving the “safer” (read: harder to get wrong) things to their own models and then the more “creative” stuff to an explicit external model, they can shift blame: “Hey, we didn’t made up that information, we explicitly said that was ChatGPT”. I don’t think they’ll say it outright like that. Because they won’t have to.