Page 450 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

As new legal documents are acquired or existing ones updated, the system automatically
index or re-index them, ensuring that the underlying data used for analysis and search remains
current and readily accessible

To ensure the relevance and accuracy of the information in RiMAP, the system relies on a robust
process of human review and selection. While AI automates the initial data collection, human
experts review and select the most relevant and up-to-date information that serve to produce
content. This hybrid approach combines the speed and efficiency of AI with the accuracy and
contextual understanding provided by human oversight, ensuring that only the most pertinent
legal data is included in the platform.

Once users pre-select legal documentation and interact with the Assistant, they evaluate
the output against four criteria (acceptability, correctness, relevance and appropriateness).
The evaluation of the user's input is used to re-instruct (prompt) the Assistant to improve the
expected output, in a continuous optimization loop, until the user is satisfied with the response.

By linking to external public legal databases, the VLA ensures that the information within RiMAP
remains continuously updated, allowing users to access the most current and reliable legal
resources.

Impact and benefits of the solution

The VLA will significantly reduce the time and effort required for legal research and analysis by
automating the collection and summarization of information from multiple external sources. By
centralizing and indexing legal data, the VLA will allow users to access accurate and up-to-date
legal knowledge quickly and maintain the quality of the legal analysis, as all the text generated
is referenced to the original documents’ official source.

The human review process integrated into the VLA will enhance the relevance and quality of
the information, ensuring that RiMAP’s content reflects the most reliable and accurate legal
data available. This approach will also reduce the time UNHCR staff spend on maintaining the
platform, improving operational efficiency.

The VLA chatbot will offer users an intuitive interface that allows them to interact with the VLA,
ask questions, and receive legal information in a more conversational and accessible format.
The VLA’s capabilities will span all 193 UN member states, ensuring users have access to a
comprehensive, global legal resource. This will empower users to navigate complex legal
systems and better understand the rights of displaced and stateless persons, enabling more
informed advocacy and facilitating greater legal support for displaced populations.

Partners

UNHCR published a Call for Expression of Interest to gradually build a legal network of partners
that will contribute to the review and analysis of the legal data collected for RiMAP. They will
be using the VLA to enhance efficiency to produce legal analysis.

2�2 Benefits of the Use Case

The combination of AI-driven technology and human oversight will enable RiMAP to remain
a comprehensive, accurate, and continuously updated resource, supporting global efforts to

414