There is a danger in oversimplify AI results to present them in a way that implies humanness/intelligence, especially when more reasonable explanations are wort

There is a danger in oversimplify AI results to present them in a way that implies humanness/intelligence, especially when more reasonable explanations are worthy of discussion. This type of writing leads to readers' hype and misconception of AI. AI performing better in the study's context shouldn't be considered a sign of intelligence. Humans which are the source of the training material write more accurately and concisely in those contexts. Rather than speculating humanness, the technical achievement is producing an AI that has this much nuance. We can then speculate what other trends are present beyond urgency, that we can exploit for enhanced results, along with how we can better tune models to reduce noise.