I'd argue that the best time to be sounding the alarm is before the tech has reached the point of causing real harm. Obviously machine translation of asylum ca

I'd argue that the best time to be sounding the alarm is before the tech has reached the point of causing real harm. Obviously machine translation of asylum cases sounds like a major issue, and becomes ammunition to tell the story of how problematic things could become with continued development. I also suspect arguments will be made like "more cases can be processed this way", which may be true on some time horizons. This isn't a zero sum game, and if anything, this renewed focus on the dangers of emerging AI tech provides an opportunity to surface situations that are already broken / in danger of harming people. (I wasn't aware of machine translation for asylum cases, so this article put that on my radar, indirectly).