EffectuationUnderUncertainty-EuroMed2026 v3FinSub
The paper re‑examines Sarasvathy’s effectuation logic for organizational decision‑making in contexts where generative artificial intelligence (GenAI) participates in problem framing, option generation, and evaluation. Using a conceptual theory‑adaptation approach, the authors first make explicit the five classic effectuation principles (means‑driven action, affordable loss, strategic partnership, leveraging contingencies, and control over prediction) and the human‑centered assumptions that underlie them. They then identify how GenAI violates these assumptions by expanding the pool of means, shifting cognitive effort from generation to evaluation, and obscuring accountability. To reconcile effectuation with GenAI, the paper introduces decision boundaries – organizational rules that govern admissibility of AI outputs, escalation thresholds, override rights, and accountability assignments. A Decision‑Boundary Process Framework is built, distinguishing three ideal‑type configurations: (i) augmented (human retains final authority, AI supports generation/evaluation), (ii) hybrid (evaluation authority shared between human and AI), and (iii) delegated (AI operates within bounded parameters, human focuses on boundary design and oversight). Six research propositions (RP1‑RP6) link each configuration to observable changes in means, affordable loss, partnership, control, cognitive‑effort allocation, and accountability. The authors argue that, when decision boundaries are deliberately designed, effectuation’s core non‑predictive control logic remains viable despite GenAI’s involvement. The paper concludes with managerial implications (designing admissibility rules, escalation mechanisms, and clear accountability matrices), acknowledges limitations (no model‑specific analysis, no regulatory/ethical discussion), and outlines future research avenues on empirical validation of the configurations and boundary drift over time.
Topics
Re‑specifying Effectuation for GenAI‑augmented Decision Contexts
Articulates how the classic effectuation principles must be revised when generative AI contributes to framing, option generation, and evaluation.