From the 'Conclusions' of Deaton and Cartwright (2018) [1], which is the one that seemingly comes up in debates among friends most often when this subject arises. I suggest to read that entire part of this paper at least -- it is very sobering. E.g.:
"[...] assumptions that RCTs claim to (but do not) avoid, so that if the aim is to use empirical evidence, any credibility advantage that RCTs have in estimation is no longer operative. And because RCTs tell us so little about why results happen, they have a disadvantage over studies that use a wider range of prior information and data to help nail down mechanisms. [...] Without knowing why things happen and why people do things, we run the risk of worthless casual (‘fairy story’) causal theorizing and have given up on one of the central tasks of economics and other social sciences."
I do not doubt everything in [1] applies even more so to RCTs in something like psychotherapy where the 'success' of 'failure' rests mostly on the interpretation by the subject and the 'why' is a black box. Until we understand and can observe brains in action much, much better and with vastly more detail than we do today that is.