I wouldn't put too much stock into the take that high skilled workers benefit the most. I see it as just a current snapshot of the research. After models exceeding ChatGPT-3.5 came out the consensus seems to have flipped.
Maybe the current state is durable, but it's by no means guaranteed.
If as the accelerationists believe AI really does commodity analytical intelligence and problem solving ability, than perhaps the lab assistant who is good at physically handling the equipment might benefit relatively more than the high skilled researcher.