That needs some clarification. Most people cannot distguish between a risk being somewhat low, and it being extremely low, so we need to be careful about the transition from numbers to qualitative divisions. The risks of being killed in a plane crash are so low that unless you're a pilot, you should ignore them; and overestimating the risks of flying would cause too much driving, which is more dangerous. In the case of AIDS, the probability of transmission may be "low", but none of the numbers given are so low that they would justify skipping any of the common safety precautions, so we shouldn't describe the probabilities involved as low in the presence of people who can't do the utility computations themselves.