You could argue that it is poorly made but perhaps the model was trained in a rush because it was a cheap, knock-off produced en-masse for school-scale.
Even if its well-made; imagine you're some 1 in a 1,000,000 non-typical. Who is the school going to believe, some irksome child or the system that works perfectly fine for 99.99% of other students? "Obviously" in such a case, the child is lying and just needs to learn to pay attention.
On the other hand it can be used to detect micro sleep in drivers and there it can save lives.
So yeah, as with any tech you can use it for good or bad and you should not use it for everything
Regardless these long-haul issues should be resolved by automated driving anyway given that long highway journeys are a decent use-case for automation.
I figure if I'm doing an illegal shift my boss has pressured me into I'm going to take the hat off because either my boss tells me to or its uncomfortable over the course of a 20 hour shift. I don't think it solves the problem in the way they pretend it does.
Nor do I buy the argument that copying what Chinese firms are doing (the example of the rail company) is necessarily a strong reason, given we might be just following micro-managing idiots into a dead-end of their own construction.
The true horror is this monstrosity is the oppression of neuro-divergent individuals by automated models trained on typical minds following happy paths.