Do you know anything about the basics of scientific rigor, informed consent and non-maleficence?
The exercise breaks all of them. Students didn't give consent. Students are hand-picked. Students were subjected to ridicule. Students potentially having psychological or emotional traumatization later on was not planned for or addressed. Statistical evidence for the exercise actually reducing prejudice is moderate, not as wildly as successful as you might think. Evidence of it actually having a useful impact towards racist perceptions is slim anyways since it's so divorced from race.
It's an interesting and important exercise, just like Little Albert, Milgram's shocks, Seligman's dogs, Harlow's monkeys and the Stanford Prison. It's not something to be admired or used on any kind of broad scale to educate about racism though. There are far-better anti-racism tools out there that have been promoted instead, and for good reason.