Symbol grounding for generative AI: lessons learned from interpretive ABM

Martin Neumann ^1* Vanessa Dirksen ²

• ¹ University of Southern Denmark, Odense, Denmark
• ² Open University of the Netherlands, Heerlen, Netherlands

This perspectives article argues that not only humanities benefit and are transformed by recent developments in AI but that also AI might benefit from the humanities. This is demonstrated with regard to the symbol grounding problem in AI, by taking into consideration that meaning is not the outcome of a two-way relation between an object and a brain (or AI), but of the negotiation of meaning in the triadic relation between objects, symbols and human practices. This is common in the interpretive social research tradition of the humanities. As we argue, AI benefits from embedding generative methods in interpretive social research methodologies. How this can be achieved is demonstrated with the example of the recently developed methodology of interpretive ABM (iABM). This methodology enables the generation of counterfactual narratives which are anchored in ethnographic evidence and hermeneutically interpreted, hence, producing symbolically grounded and plausible futures. Criteria for plausibility correspond to contemporary guidelines for the assessment of trustworthy AI, namely human agency and oversight, transparency and auditability.