The UK Department for Transport (DfT) has launched the Consultation Analysis Tool (CAT) to expedite the analysis of citizen feedback from public consultations. Each year, the DfT manages around 55 consultations, generating over 100,000 free-text responses which previously required extensive manual review over several months.
The CAT, developed in collaboration with Google Cloud and the Alan Turing Institute, utilizes Google’s Vertex AI platform and Gemini models to categorize themes from public feedback within hours, achieving up to 90% accuracy. This new capability allows the DfT to publish consultation responses within 12 weeks, significantly improving responsiveness to citizens.
The new tool has resulted in potential savings of up to £4 million annually for the DfT by streamlining the feedback process. The CAT has already been employed in analyzing public comments related to the Integrated National Transport Strategy and enhancements to driving test booking rules.
In addition to the CAT, the DfT has also implemented a Connectivity Tool, developed using Google Cloud services, to support urban planners in making sustainable infrastructure decisions. Furthermore, the AI Correspondence Drafter generates first drafts of responses to public inquiries by efficiently retrieving and processing relevant policy information.
The DfT underscores the importance of responsible AI usage, highlighting a commitment to maintaining human oversight in policy decisions. Employing a “human-in-the-loop” model, the department ensures that AI outputs are scrutinized for accuracy, fairness, and bias, aligning with public interest and expectations.
While Google Cloud provides the necessary processing power for these tools, final decisions are made by DfT policy experts. This strategy aims to enhance the efficiency of the UK’s transport system while prioritizing accountability in the use of AI technology.
