Transparency and legal adaptability in policymaking
During the final round of testing the prototype with policymakers, they confirmed that the tool facilitates informed decision-making by improving the decision-making transparency and ensuring that diverse stakeholder perspectives are considered during the policymaking process. Nevertheless, legal concerns about the tool’s adaptability were raised by the policymakers of the Ministry.
Experts expressed concerns that the tool might silo certain perspectives, particularly those of human stakeholders.
Emphasising the need for objectivity, experts recommended including all human stakeholders in the lens setup and giving them a voice in the dashboard.
Based on feedback from an expert about users and roles, the identification of a user has been added to define the hierarchical role of the user in the system.
0. Set up a ‘perspective’ lens for the AI model to analyze uploaded draft documents.
1. Upload any policy-related files
A summary of the uploaded files has been added to contextualise the AI model's interpretation of the uploaded files as feedback to the user.
2. Dashboard with response of lens and data visualisations (based on the uploaded files)
Experts noted that, to build trust, transparency should go beyond search history by offering a comprehensive decision-making overview to improve communication between policymakers and stakeholders. Therefore, a 'change log' has been introduced.
More detail has been added to the explanation of the AI confidence indicator through human assessment.
3. Guidelines for improvements based on the lens and selected data visualisation scenario
The sense of control and autonomy, based on user feedback, has been included by allowing users to edit the suggested guidelines generated based on the (previous) the refinement parameters.
AI Discosure
A graphic representation of the model’s functionality has been added, along with a human explanation, to enhance transparency based on feedback from the experts.