Generative AI–Powered Road Weather Drive Assist System Received 88 Percent User Satisfaction for Ease of Use and Actionable Insights Based on Eight Users’ Feedback.

A Survey-Based Evaluation Assessed the Usability, Value, and Integration Potential of a Conversational Artificial Intelligence Based Road Weather and Information System (CARWIS).

Date Posted
02/26/2026
Identifier
2026-B02031

An Intelligent Human-Centric Communication System for Adverse Weather and Road Conditions

Summary Information

Hazardous events such as snow, ice, and low visibility significantly elevate crash risk and impair drivers’ ability to anticipate and respond to rapidly changing roadway environments. This study developed a Conversational Artificial Intelligence (AI) for Road Weather and Information Systems (CARWIS) that delivered real-time, easy-to-understand information on Iowa’s road and weather conditions to drivers. To assess the usability, effectiveness, and relevance of CARWIS, a survey was developed and administered to transportation professionals and domain experts, following their direct use of the system.

METHODOLOGY

CARWIS interface was composed of four specialized AI experts, including the Drive Assistant, Winter Expert, Live View Expert, and 511 Data Expert, designed to deliver targeted, real-time road and weather information to drivers. The interface supported real-time conversation, displaying system responses and camera images based on the users’ request. The system used Iowa 511 winter road conditions and events data, snowplow truck location data, and images from traffic cameras, and integrated them with a large-language model to process and respond to natural language user queries. 

A survey was conducted to assess CARWIS, in which the questions were grouped into three: general system usability and value, agent-specific performance, and cross-agent comparison with forward planning. Eight valid responses from transportation professionals and domain experts were gathered and analyzed.

FINDINGS

  • Overall, the survey results showed that CARWIS delivered clear benefits in usability, practical value, and integration potential.
  • Seven out of eight participants (88 percent) agreed or strongly agreed that CARWIS was easy to use, indicating that the system’s interface and interaction design was intuitive and accessible. 
  • Half respondents agreed or strongly agreed that CARWIS provided clear and understandable information, which suggested that the system generally communicated information effectively.
  • Five of eight respondents agreed or strongly agreed that CARWIS aligned well with the existing DOT workflows, and it was generally compatible with current operational practices without requiring major procedural changes.
  • Seven of eight respondents agreed that CARWIS provided value beyond tools currently available at their agencies, indicating that the platform introduced novelty.
  • Six respondents agreed and one strongly agreed that CARWIS features could be integrated into the existing DOT systems such as Iowa 511, highlighting strong perceived potential for complementary deployment rather than replacement.
  • Seven of eight users reported that the Drive Assistant provided actionable and trustworthy information, and all users rated live traffic camera views from the Live View Expert as very helpful.

An Intelligent Human-Centric Communication System for Adverse Weather and Road Conditions

An Intelligent Human-Centric Communication System for Adverse Weather and Road Conditions
Source Publication Date
09/01/2025
Author
Demir, Ibrahim; Yusuf Sermet, Moaiz Abrar, and Gabriel Vald
Publisher
Prepared by IIHR-Hydroscience & Engineering Laboratory at the University of Iowa for the Iowa Department of Transportation and U.S. DOT Federal Highway Administration
Other Reference Number
Aurora Project 2023-02
Results Type
Deployment Locations