When asking about the future of technolinguistic tools, like speech to text or predictive smart replies, we need to ask not just how they can be used, but how they can be subverted; not just how designers can help users communicate their intentions, but how users can help them communicate more than the designers intended.
