Satisficing vs. Maximizing in Prompt Writing: Trait and Task Effects in Human–AI Interaction

Paper Title

Satisficing vs. Maximizing in Prompt Writing: Trait and Task Effects in Human–AI Interaction

Publication Info

• Topic area: Psychological factors influencing human–AI interaction in prompt writing.
• Keywords: Generative AI, satisficing, maximizing, prompt engineering, human–AI interaction, cognitive styles, decision-making, task context, behavioral psychology, distributed cognition.

Background and Problem

• Problem / challenge: Limited understanding of how psychological factors, such as satisficing tendencies, influence prompt-writing strategies when interacting with generative AI systems.
• Significance: Insights into these behaviors can improve the design of AI systems, enhance task performance, and reduce cognitive effort in AI-assisted work.
• Motivation and related work: Previous research has focused on technical aspects of prompts or user attitudes toward AI but has not extensively explored the cognitive decision strategies underlying prompt formulation. This study bridges behavioral psychology and human–AI interaction to address this gap.

Solution

• Proposed approach: An empirical study investigating how individual satisficing tendencies influence maximizing behavior in prompt writing across different task scenarios.
• Novelty:
  1. Introduces the satisficing-maximizing framework to explain differences in prompt-writing behavior.
  2. Examines the role of individual traits (satisficing tendency) and task characteristics in shaping prompt strategies.
  3. Provides practical recommendations for designing adaptive GenAI systems.
  4. Offers the first empirical study on satisficing and maximizing in the context of human–GenAI interaction.
• Procedure and key techniques:
  • Conducted an online vignette experiment with 132 participants.
  • Measured satisficing tendencies using the Short Maximization Inventory (SMI) and included controls for algorithm aversion and prompt-writing competence.
  • Participants chose between satisficing and maximizing prompt options across five task scenarios: Job Offer Choice, Creativity Support, Research Support, Writing Feedback, and Technical Support.
  • Analyzed data using linear mixed models (LMM) and ANCOVA to assess the influence of individual traits and task context on maximizing behavior.

Results

• Concrete findings:
  • Higher satisficing tendencies were associated with reduced maximizing behavior (β = −0.332, p = .018).
  • Higher self-reported prompt-writing competence predicted increased maximizing behavior (β = 0.253, p < .001).
  • Task type significantly influenced maximizing behavior, with more maximizing observed in Job Offer Choice and Creativity Support scenarios and more satisficing in Writing Feedback and Technical Support tasks.
• Advantage over baselines:
  • Demonstrates that psychological traits (satisficing) and task context systematically influence prompt-writing strategies, extending beyond prior studies focused solely on technical or situational factors.
• Experiments / evaluation:
  • Sample size: 132 participants.
  • Measures: SMI, Algorithm Aversion Scale, self-reported prompt-writing competence.
  • Statistical methods: LMM, ANCOVA, and post-hoc analyses.
  • Scenarios designed to reflect varying levels of cognitive effort and task stakes.
• Limitations and future work:
  • Vignette design constrained participants to pre-defined prompts, limiting naturalistic behavior.
  • Did not measure perceptions of AI competence or trust.
  • Sample limited to English-speaking participants; cultural and occupational norms may influence results.
  • Future research should include real-world interaction logs, iterative prompt-writing tasks, and broader task domains like healthcare or law.

Summary

This study investigates how individual satisficing tendencies and task characteristics influence maximizing behavior in prompt writing for generative AI systems. Results show that stronger satisficing tendencies reduce maximizing behavior, while higher prompt-writing competence increases it. Task context also shapes strategies, with high-stakes or creative tasks eliciting more maximizing behavior. These findings extend theories of bounded rationality to human–GenAI interaction and suggest that adaptive AI systems should tailor support based on user tendencies and task demands. Future work should explore real-world interactions and broader application domains.