Making FAIR trustworthy

The life sciences community is now increasingly leaning on the processing powers of machines to carry out advanced scientific research. So in order to adequately exploit the capabilities of machines in science, the FAIR (Findable, Accessible, Interoperable, Reusable) principles for scientific data management and stewardship have been proposed. These principles are to assist scientists in tweaking their established research routines so as to unlock the true potential of machines and contribute to better science. However, through interviews with key scientist groups implicated by FAIR we have learned that doing what FAIR demands also presents certain epistemic concerns that discourage scientists to trust FAIR. One such concern is the loss of valuable knowledge in the translation of versatile human readable research output to a restricted, machine friendly language to enable machine action (semantic freedom). A second concern is evident in the displacement of human expertise by this increasing reliance on machines and the ensuing loss of knowledge contributed by these displaced experts (expert intuition). Thus, through this article, we highlight how incorporating FAIR also presents an epistemic loss to the scientific community. But the lack of attention given to these concerns by the proponents of FAIR offers scientists who have to abide by FAIR sufficient reason to resist it. We thus propose that while the implementation of FAIR has so far been paternalistic and top-down, such concerns have also made the scientist sceptical. So by initiating a more balanced, open and honest discussion of not just the benefits and promises of FAIR but also such epistemic costs, FAIR could lay to rest reasons for such scepticism and foster trust within the stakeholders of FAIR.