They’re Large Language Models. They’re defined as generative pre-trained text transformers, that’s their entire purpose.

Saying the calculator spits out random numbers would be wrong, but saying a calculator spits out numbers, that would be correct. Reductionist would probably be a better word than regressive or asinine.

But it’s accurate? Doesn’t mean that human looking text can’t be helpful to some, but it’ll also help keep us grounded to the reality of the tech.

I don’t know the details but this feels like such a specific attack vector. Most malware targets the easiest and most common payload delivery mechanism as possible. Having someone connected via hotspot and piggybacking ontop of a specific workflow such as Shizuku just seems super unlikely. Could absolutely be wrong about this though, just my gut feel