The UK Department for Transport (DfT) has developed a Consultation Analysis Tool (CAT) in partnership with Google Cloud and the Alan Turing Institute to enhance the analysis of public feedback from consultations.
The tool addresses the challenge of managing approximately 55 public consultations annually, which generate over 100,000 free-text responses. Historically, analyzing this feedback required months of manual review. The DfT aims to publish consultation responses within 12 weeks, significantly speeding up the feedback process.
The CAT, built on Google’s Vertex AI platform, utilizes Gemini models to categorize themes from large volumes of feedback in just a few hours, achieving up to 90% analysis accuracy. This efficiency allows the government to respond more quickly to citizens and is projected to save the DfT up to £4 million annually by 2025.
One application of the CAT includes analyzing public comments related to the Integrated National Transport Strategy. Additionally, the DfT has created a Connectivity Tool with Google Cloud services to assist urban planners in making sustainable infrastructure decisions.
The DfT also introduced an AI Correspondence Drafter that generates initial drafts for public inquiries by leveraging Vertex AI Search and Gemini for drafting. The use of AI aims to transform transport efficiency while ensuring responsible implementation with human oversight.
By employing a “human-in-the-loop” model, the DfT ensures that AI outputs undergo checks for accuracy, fairness, and bias. Research indicates that the public values these factors in AI outputs. Google Cloud supplies the computational power, while final policy decisions rest with DfT experts.
This approach aims to ensure that technological resources serve the public interest and improve the nation’s transport system.
