Compared to those given no messages, these participants produced more stereotypical ratings, whether about women, older people or the obese.
It would be more interesting to measure the correctness of the ratings. A stereotype, unlike some definitions of "bias", is not automatically wrong; it could just as well be correct. "Men are physically stronger than women" is a stereotype which is correct and useful (the difference has a significant magnitude).
This formulation of evidence completely disregards an important factor of bayesian probability which is that new evidence incrementally updates your prior based on the predictive weight of the new information. New evidence doesn't completely eradicate the existence of the prior. Individual facts do not screen off demographic facts, they are supplementary facts that update our probability estimate in a different direction.
It looks like telling people "everyone is biased" might make people not want to change their behavior to overcome their biases:
The authors suggest that telling participants that everyone is biased makes being biased seem like not much of a big deal. If everyone is doing it, then it's not wrong for me to do it as well. However, it looks like the solution to the problem presented here is to give a little white lie that will prompt people to overcome their biases: