> EDIT: Most papers/presentations already talk about the possible positive impacts -- nobody holds back on that for lack of expertise. That's not really accura

> EDIT: Most papers/presentations already talk about the possible positive impacts -- nobody holds back on that for lack of expertise. That's not really accurate. Sure, people talk about positive impacts - but in a technical CS paper, they talk about positive technical aspects, e.g. "this new method will allow you to train ML models faster". They don't usually talk about societal impacts, which is what the article is about.