Many years ago, when I first read Steven Pinker’s book The Language Instinct (highly recommended), I learned that natural language is imprecise for a very good reason. As language evolved it gave us an evolutionary advantage as a species.
We evolved spoken language primarily to build networks of trust with our fellow humans, and that’s a kind of super power. Before we know whether we can trust somebody, we are able to speak to them in imprecise ways, so that our meaning and intention is able to remain fuzzy.
This gives us plausible deniability. And that lets us back away from interpersonal connections gracefully as needed.
So it may seem at first as though natural language, as opposed to, say, a language used for computer programming, is bad at conveying exact meanings. But in fact it’s quite the opposite: Natural language is excellent at precisely adjusting intentional imprecision.
I wonder whether it would be useful to design a computer programming language that is focused specifically on this kind of intentional fuzziness. As artificial intelligence takes on more and more human-related tasks, that might be a fruitful direction for language design research.