Travelers hand more planning power to AI when systems are dependable and accountable, but too much explanation backfires; trust matters as a baseline for low-level tasks while excessive transparency deters delegation.

Main Finding

Generative AI is reshaping travelers' delegation of planning decisions. Among five responsible-AI signals (reliability, fairness, trustworthiness, accountability, transparency), reliability and accountability consistently increase willingness to delegate across all delegation levels (attribute set, choice set, final decision). Trustworthiness acts as a threshold condition (important mainly at the attribute level). Excessive transparency can produce cognitive overload and reduce willingness to delegate to GenAI.

Key Points

• Context & theory: The paper frames traveler delegation to GenAI using agency theory, trust theory, and a decision-delegation framework.
• Delegation levels:
  • Attribute set — delegating specification/weighting of attributes (e.g., desired amenities).
  • Choice set — delegating generation/filtering of candidate options.
  • Final decision — delegating the actual booking/selection.
• Responsible-AI characteristics examined: reliability, fairness, trustworthiness, accountability, transparency.
• Core empirical results:
  • Reliability and accountability are robust, positive drivers of delegation across all three levels.
  • Trustworthiness functions as a threshold (significant at the attribute level).
  • High transparency can backfire: too much explainability/complexity causes cognitive overload and lowers delegation willingness.
• Practical takeaways: designers should prioritize dependable, accountable GenAI systems and calibrate transparency to avoid overwhelming users.

Data & Methods

• Empirical basis: multi-method approach using data from 421 travellers.
• Variables: measures of perceived reliability, fairness, trustworthiness, accountability, transparency, and willingness to delegate at three delegation levels.
• Analysis: the study tests how the five responsible-AI signals predict delegation across levels (the summary does not specify the exact statistical techniques; the paper uses mixed quantitative/qualitative methods consistent with a multi-method design).

Implications for AI Economics

• Adoption and market dynamics:
  • Systems emphasizing reliability and accountability should see higher adoption and deeper delegation, increasing demand for GenAI travel services and raising incentives for firms to invest in those properties.
  • Trustworthiness acting as a threshold implies segmented demand: users may accept higher automation for low-level tasks only after basic trust criteria are met.
• Product design and competition:
  • Firms face trade-offs between explainability and usability. Overly detailed transparency can reduce adoption—suggesting a market for tiered explainability (simple summaries for most users, richer explanations on demand).
  • Providers who credibly signal accountability (auditability, redress mechanisms) can capture more users and command premiums or market share.
• Labor and intermediaries:
  • Increased delegation at attribute/choice levels could substitute routine planning work (reducing demand for some intermediary services) while shifting value to higher-touch human advisors and oversight roles.
• Policy and regulation:
  • Regulators should prioritize accountability standards and clarity around responsible-AI claims; transparency mandates should be balanced with usability considerations to avoid unintended welfare losses from cognitive overload.
• Research directions for AI economics:
  • Quantify welfare effects of delegation across levels (consumer surplus vs. cognitive costs).
  • Model market equilibria where firms choose transparency levels and accountability investments given heterogeneous consumer delegation thresholds.
  • Study heterogeneity in delegation (by experience, risk preferences) to predict adoption and labor impacts across segments.