AI models that consider user's feeling are more likely to make errorsarstechnica.com2 points·milkglass··0 commentsOpen articleSaveView on HN