Many of us have had the experience of sitting through college courses taught by grad students. For my part, the results were extremely mixed. I had a few grad student instructors who were pretty good, many who were mediocre, and a few who were atrocious teachers.
Apparently my experience wasn't unique, judging by the attention received by a recent paper published in the journal Science. A review of it appearing in the Chronicle of Higher Education says "Trained but inexperienced postdoctoral students can teach a college class as well as or better than longtime professors who rely on lectures, if the postdocs learn to incorporate a method of teaching that relies on having students interact with the material they are learning through discussions and assignments synthesizing new and old information and experiences."
As you read through the C of HE review it becomes clear they are talking about directed learning techniques that we highlighted in an editorial three years ago (http://machinedesign.com/article/leland-teschlers-editorial-not-that-hard-to-learn-1009). A school called the Morningside Academy in Seattle uses these techniques to get not just kids but also nursing students and other pre-professionals to learn material at a high rate of speed. In one case, academically deficient adults improved by two academic grade levels after about 18 hours of Morningside-inspired coaching.
As the people at Morningside told me, it takes a motivated and engaged teacher to actually use these techniques effectively. And that is the rub with unleashing them on grad student instructors. Not of a few of the instructors I had seemed preoccupied in class or just gave the appearance of wishing they were somewhere else.
So I am afraid physics undergrads are safe from exposure to these effective teaching techniques. A lot of those posting comments to the C of HE item seem to agree: http://chronicle.com/article/Postdocs-Can-Be-Trained-to-Be/127525/?sid=at&utm_source=at&utm_medium=en