I don't think anyone doubts that often people will spend additional money they are given productively.
I'm not trying to be unreasonably judgemental, and I don't mean things like bond rates or yield curves. Things like making sure you have enough dollars at the end of the month to make a rent payment. Making sure the check gets written so it actually happens. Keeping track of automated withdrawals and ensuring the number stays above zero with the timing of all of them.
This is not trivial at all for a huge chunk of the population that is served by social welfare programs. You can't just expect it to work out giving them cash instead.
The vast majority of poor people do just fine with money when given to them. The 1-2% who are addicted to substances or mentally ill will always need additional services, but why throw out the huge efficiency gains in just direct cash transfers to all because of the worry that 1-2% might waste it? The efficiency gains of a UBI vs. most means-tested programs is likely more on the order of 20-50%.
A similar argument is “would we give a UBI to billionaires too?! What a waste!” Yes, what a waste of 0.0000003% of the UBI!
One shortcoming of the UBI studies I've heard about is that (naturally) they focus on settings that I don't have much confidence would necessarily translate to other settings like the current US. I can easily imagine the efficiency of UBI changing over time, let alone across locations, due to a plethora of factors. This isn't really a criticism of the studies (they're probably doing the best they can, and achieving good results on those is obviously a prerequisite), but rather of how much we can extrapolate from them and how much weight we can put in positive results currently. No matter how perfect their results, I'd totally expect people to be reasonably skeptical of their extrapolability (if that's a word) in the beginning.