You already have an (implicit) prior probability distribution over future events without exact precedent, roughly proportional to how much you'd be surprised to see any of them happening.

If someone offered you the bet (at even odds) that a fully automated (Sebastian Thrun-style) highway would be developed before human-cat hybrids were developed, I think you might well take that bet. That's a statement of your probability distribution.

These probabilities are much harder to calibrate than probabilities of frequencies of random events, but the important thing is that you already have rough estimates for them.