Stefano Canali, researcher in philosophy of science at the Department of Electronics, Information and Bioengineering of the Politecnico di Milano whose work is supported by the NRRP NextGeneration EU project FAIR, has published a letter in Nature on the use of Artificial Intelligence (AI) for evidence synthesis and evidence-based policy, in interdisciplinary collaboration with Francesco Barone-Adesi from the University of Eastern Piedmont.
The letter argues that, while AI has great potential for informing policy, we should not think that it can create judgement-free policy advice. Systems based on AI are not free of the influence of judgements, values, and bias and their black-box nature can actually make this influence more opaque and difficult to scrutinise. Seeing AI as something neutral that is free of values and judgements is potentially dangerous, particularly when results have an impact on policy-making and governments might be tempted to claim that their decisions can be neutral because they “follow the science”.