I'm not going to write a large detailed literature review, but my own biases against intelligence psychology come from two points.
First, it's reductionist. By further defining generational intelligence, they borrow models from cognitive psychology in determining mental ability using abstract measures like perceptual speed, spatial visualization, word fluency, etc. and then creating overall measurement in the form of an intelligence quotient. There is no theory or interactive concept behind it, it relies on largely descriptive models in cognitive science and neuroscience, applies normative value (what is positive and negative ability), finding correlation between these now normative structures and if those models and theories change, then by nature it has to fundamentally reassess itself because it is a reductionist science.
My second point is its lack of causation research. Intrinsically speaking, general intelligence psychology is largely correlational and lacks solid causation research as most causation research is shaky and have conflicting evidence, aside from processing speed which seems to be the most promising causal factor:
This isn't usually a problem, but intelligence psychology is again normative. It applies good and bad measure to ability but there can't be attempts to increase it. In fact according to most studies apart from some developmental studies (which again is largely correlational and predictive), you can not increase IQ.
My extra third point that I found googling is reification, which I think is also a solid criticism. This criticism states that it makes concrete what is significantly abstract. This can be seen from an anti-psychology perspective to its use of mathematical analysis.
Personally I'd liken intelligence psychology to something like personality psychology. Both attempt to apply normative variables to certain cognitive processes and try to determine relationships with these normative values and external ones. All of this leads to me thinking IQ as a tool may be useful information for correlation analysis and research, but again I don't think it's useful for value judgements or causal determiners. If you want to understand your cognitive processes, get into a field whose latent variable status isn't questioned every decade.