I actually sometimes like to do polls, but only if a real human is on the line. I think the polls may skew quite a bit in odd and unpredicted ways if they are using robots to call. They are selecting for a very specific bunch of people who will sit through robot calls. I think that select group may be different from the general population.

Right now we are having front page news in my area about how two leading polling companies have such different results from polling for our senate race. One shows candidate 1 leading by a slim margin. The other shows candidate 2 leading by a huge margin. There is no overlap even considering the polls published errors of margin. Perhaps one company is using the robo-calls and the other is having humans call.