Should I Ask? AI and the Decision to Negotiate

Abstract: In many workplace contexts, especially for early-career professionals, expectations around negotiability are ambiguous, and individuals have to infer whether initiating negotiation is appropriate in the absence of clear norms. Building on work on weak situations and negotiation initiation, we argue that in ambiguous settings, anxiety and low negotiation self-efficacy can operate as barriers to initiating negotiation. We propose that access to large language models (LLMs) may reduce these barriers by providing normative clarity (“Is this negotiable?”), informational clarity (benchmarks, options), and procedural clarity (how to phrase a request, how to sequence a conversation, how to respond to resistance). We plan to test this argument in an experiment that manipulates access to LLM-based decision support in a workplace scenario where negotiation is possible but not explicitly prescribed. We will examine whether LLM access increases negotiation initiation through reduced anxiety and increased negotiator self-efficacy.

Keywords: negotiation initiation, ambiguity, LLMs, anxiety, negotiator self-efficacy