Skip to main content
Preprint Credible — Major labs, established NGOs, reputable named-author preprints

Annie, Are You Okay? How Style- and Context-Based Personalization Shape AI-Assisted Decision-Making

Preregistered 2 x 2 between-subjects experiment (N=240) in which participants ranked three comparably viable stocks, discussed them with a GPT-5.4-based assistant whose personalisation was varied by communication style and by use of the participant's stated context, then re-ranked. Only context-based personalisation reliably changed behaviour, increasing reconsideration and moving rankings toward the assistant's assigned recommendation, while participants did not rate the assistant as more correct or trustworthy.

Publisher

arXiv (Northeastern University; University College Dublin)

Published

21 Sept 2026

Added

today

DOI

Key Findings

  • Context-based personalisation increased reconsideration and moved rankings toward the AI's assigned recommendation; style-based personalisation was perceived but did not reliably change rankings
  • Participants reported feeling more influenced without judging the AI more correct, trustworthy, intelligent, likeable or higher quality
  • Participants initially farther from the recommendation moved more toward it while rating its advice as less correct
  • Exploratory analyses suggest greater susceptibility among lower-expertise participants
  • All four cells used GPT-5.4 with prompt layers implementing the personalisation manipulations

Methodology Notes

IRB-approved, preregistered 2 x 2 between-subjects design, N=240 online participants; stock-ranking task with defensible options and no ground truth; GPT-5.4 in all cells. Motivated by ChatGPT's June 2026 personal-finance feature. v1 posted 2026-09-21. Self-report outcomes; single task domain.

Authors

Hasibur Rahman, Benjamin R. Cowan, Smit Desai

Tags

personalizationfinancial-advicepreregisteredgpt-5.4decision-support

Cite This

APA

Hasibur Rahman, Benjamin R. Cowan, Smit Desai. (2026). Annie, Are You Okay? How Style- and Context-Based Personalization Shape AI-Assisted Decision-Making. arXiv (Northeastern University; University College Dublin). https://arxiv.org/abs/2609.24644