AI in industry was basically made my PhD grads. Without that pipeline, there would be no AI, and I am not exaggerating much at all.
I think a mix of the current system with more permanent researchers makes sense.
There is a lot of work in research that fits the permanent worker better than the fresh 22 year old. But having that fresh talent is really beneficial to science.
> If students decide that they are comfortable with a sub 20% job placement rate, then great, nothing needs to change.
The problem is in my opinion not this low job placement rate per se (it is very easy to find out that this is the case for basically every prospective researcher). The problem rather is the "politics" involved in filling these positions, and additionally the fact that positions are commonly filled by what is currently "fashionable". If you, for some (often good) reason, did good research in an area that simply did not become "fashionable": good luck finding an academic position.
The current information assymetry is exploitative. One of two things would happen under my proposed system: either nothing would change because students think they are getting a good deal as is or students don't think the deal is worth it which means that the current system only works because students are having the reality of the job market hidden from them.