It's like with prototypes in any other area. You have idea, you either convince someone that it's good (typically with alredy working prototype, which precludes funding for prototypes) or use your own time/funds. I wish there could be other way. Maybe someday UBI will give us that.
If you did not have to consider financial viability of a project, then more people would make more random stuff, which of course would produce more useful stuff, because the sample size increases, but there would also be more useless garbage.
The question is whether and in what direction/magnitude the signal to noise ratio would change in the UBI scenario.
No, current way is the most optimal I can think of in current situation. There are better ways, but not widely implementable currently.
As for more garbage - yes, there would be more noise, but ALSO more signal (in my humble opinion). Signal in this case is useful things, which would be multiplied by usability, which is a net win for me.