Philosophers can and do criticize scientific research and findings all the time.
Philosophers of science sometimes think of scientific research as divided into "phase" or "stages." This way of thinking about research can be misleading — scientists don't actually move discretely from stage or steps, and they typically we move back and forth between different stages — but it's useful for understanding the variety of ways in which philosophers critique scientific research.
So here's one way of dividing scientific research into stages:
- Asking research questions
- Designing studies to address those questions
- Analyzing and interpreting data gathered from those studies
- Combining findings from several studies to formulate answers to the research questions
I'll run with Mauro ALLEGRANZA's example of gene editing to treat cancer.
In the first stage, philosophers can critically identify questionable assumptions or framings of research questions. For example, the idea that we can use gene editing to treat cancer might assume that cancer is "fundamentally" a molecular biology process. This is a reductionist way of thinking about cancer. We can also view cancer as a tissue-level process (the formation of cancerous tissues) or an ecological process (exposures to toxic chemicals cause a lot of cancers). Or, indeed, we can view cancer as process that unfolds on all three of these levels. These different framings might lead us to pose different research questions.
When designing studies, scientists must build a "ladder" of procedures and data, connecting theoretical concepts (cancer, genes, "gene editing") to the kinds of things we directly observe (Manhattan plots, microscope images). Here again, philosophers can critically identify questionable methodological and conceptual assumptions. For example, philosophers might argue that conventional genomics methods don't allow us discover disease mechanisms. That is, roughly, these methods give us correlation rather than causation, and so have limited clinical value. Philosophers can also point out that different researchers operationalize central concepts or phenomena in different ways.
For the third stage, a large body of recent work on "inductive risk" shows that ethical and political values play an essential role in the interpretation of data. Inductive risk directs us to consider the ethical significance of false positive and false negative errors. If we're doing a very rough, proof-of-concept gene editing study, we can probably tolerate very significant errors in our analysis and interpretation. But if we're trying to decide whether clinical trials provide sufficient evidence that a treatment is safe and effective, our standards should be much higher in general. And some kinds of errors (the treatment turns out not to be safe after all) are much more significant than others (the treatment turns out to be more effective than we thought).
Finally, with several studies completed, scientists need to synthesize the results and draw an overall conclusion. Many scientists argue that we should do this using meta-analysis. But philosophers have argued that meta-analysis is an extremely messy and contingent process, suggesting that it's much less reliable or impartial than its proponents claim. In addition, meta-analysis isn't well-suited to dealing with evidence from many different kinds of studies, and doesn't give us the kind of knowledge we need to extrapolate across cases.