ELSA 2.0

By Friso van Houdt

In the face of the new challenges presented by AI, the ‘contract between science and society' is bound to be revisited by the large-scale application of AI. Research into the ethical, legal and social aspects of cutting-edge technology is therefore crucial (Zwart & Neelis 2009), among other for strengthening the academic quality and social robustness of AI. ELSA research addresses this relationship between new and emerging technosciences in terms of co-production and reflexive co-evolution (Zwart & Neelis 2009). There are at least four important features typical of an ELSA approach: proximity, early anticipation, interactivity, and interdisciplinarity (Zwart & Neelis 2009). However, ELSA programs have been criticised for being ineffective (that is, it has little to no impact on policy making), and they have been critiqued for a pro-technology bias or as handmaiden of technology (i.e., the critique being that ELSA is used as promotion of technology in the guise of education to avoid challenges and criticism) (Zwart & Neelis 2009).

Taking on board the four elements typical of ELSA approaches (proximity, early anticipation, interactivity, and interdisciplinarity), and taking seriously the two critiques (ineffective and pro-technology bias), inspired our ELSA 2.0 research design and impact pathway: 1) societal impact: the ambition to increase policy relevance (e.g., a particular focus on a policy tool and the convergence work that needs to be done not being reduced to problem solving (solutionism) but as engaging in public issues, and 2) reflexivity, mutual exploration and participatory criticism:  our proximity as researchers to the object of study implies a continuous reflection upon pro-AI bias (close interaction leads to relevant results, yes, but also risks and downsides, so a questioning stance about how to remain critical of the promotors and developers of AI); mutual exploration (change of habits and finding a balance while allowing for frictions among the quintuple helix stakeholders), and; the serious “zero option” in all the case studies whereby “No AI” is seriously considered as outcome (‘participatory criticism’).  

With our ELSA research we intend to stay away from the often deterministic and simplistic social framework in which AI is typically cast. We aim at understanding the complexity of AI-cologies. ELSA 2.0 should replace caricature with nuance and provide a richer vocabulary for understanding the societal impacts of AI in public safety issues. This calls for a research strategy that is characterised not only by proximity, early anticipation, interactivity, and interdisciplinarity but also by a willingness to reflect continuously in a critical manner on the opportunities and challenges involved in the approach (societal impact, reflexivity, participatory criticism).