I'm sure I'll get heat for this, but I think a low-resolution first pass a la keyword search was actually an excellent call from a macro standpoint. It (1) clears out a good chunk of cruft, instantly, with little effort and (2) scares people away from doing further explicitly ideological work.
There are a surprising number of social scientists who engage in both good basic research and useless ideological bone-picking, and the low resolution pass effectively pushes them to focus on their real work.
Ah yes, nothing incentivises focus on "real work" like having to check that your abstract doesn't contain any words that trigger the administrations' censors after some ignorant edgelord's algorithm defunded your study of transgenic mice or black holes.
Luckily now the administration has great scientists like RFK Jr setting research priorities, no more useless ideological bone-picking is taking place