Text this: Symbol grounding for generative AI: lessons learned from interpretive ABM