Whereas many find the primary use case of AI to be coding, there's a lot of simple tasks that would otherwise be too expensive to attempt without it.

Translation of materials to various languages using Google translate is pretty expensive and with low quality for low-resource languages, but it is now nearly flawless with the Gemini 2.0 + 2.5 model of families, even for African languages. If we need to reach more people we need to leverage these tools to generate the languages they speak.

To add to this, I'd also say that there's not really a huge noticeable jump between M1 and M2. I'd just go for the M1. If you want brand new though, I think you can only find the M2.