I think you're downplaying the chances that a singularity does happen in my lifetime. 90% of experts seem to think it will.

The experts are biased.

Consider two competent AI researchers, Alice and Bob. They both investigate the possibility of a singularity. Alice comes to the conclusion that it might happen in a few centuries or never. Bob comes to the conclusion that it will be possible in a few decades. What happens next?

Alice isn't interested in the singularity any more and goes off to work at, I don't now, image recognition. Bob as a consequence of his views still is interested in the singularity and continues to work on it. At this point Alice is not an "expert" on singularity, but Bob is. A survey would ask for Bob's opinion but will not ask Alice what she thinks.