Skip to main content
European Commission logo
en
Strategy and policy
Reforms and Investments

Designing the automated data labelling platform for AI cases

Supporting reforms to develop well-regulated, stable and competitive financial markets

Funding Programme
Year
  • 2024

Designing the automated data labelling platform for AI cases 

Data labelling refers to the process of assigning tags to specific sections of documents, making it easier to locate and analyse relevant information. The Central Bank of Ireland oversees a large volume of complex supervisory reports, including Solvency and Financial Condition Reports. Current data labelling methods involve significant manual validation, are prone to inconsistencies, low number of reports processed, and lack a centralised labelling platform. To address this, The Central Bank of Ireland initiated a project with the support of the European Commission to evaluate data labelling vendors to identify the solution that best meets its needs, ultimately supporting more effective oversight and decision-making.

Context

As part of the project, a Proof of Concept (PoC) was conducted to evaluate the efficiency of a third-party data labelling solution. The goal was to assess its ability to improve label accuracy, reduce manual effort, improve collaboration between technical and business teams, and integrate seamlessly into the Bank’s existing environment. This initiative aligns with the European Commission’s priority to promote digital transformation and the use of AI in regulatory and supervisory processes, which promotes responsible AI adoption, and fosters regulatory efficiency through digital transformation.

Support delivered

The project was designed to tackle several critical challenges: reducing the extensive manual labour in data labelling to enable AI use cases, enhancing the quality and consistency of labelled data, and ensuring the selection of a purpose-fit solution through a robust assessment framework. Additionally, it aimed to establish a clear roadmap for integrating an external data labelling solution into the bank’s ecosystem while raising awareness about the strategic importance of data labelling in driving AI adoption. 

A key element was the execution of a Proof of Concept for a specific use case, which validated the preliminary to-be state and provided practical insights for refining the final recommendations and showcasing the value of a data labelling platform to bank executives and sponsors. This approach laid the foundation for a scalable, efficient platform integration that supports AI-driven decision-making and positions the organization for sustained innovation.

Achieved results

  • A comprehensive evaluation, As-Is, and roadmap for implementing a scalable data labelling platform To-Be state, enabling the adoption of a Data Labelling platform to advance AI adoption within the bank.
  • Clear vendor recommendations based on a structured assessment, ensuring alignment with CBI’s needs.
  • Validation of the To-Be state through a PoC, demonstrating feasibility for a specific use case.
  • A refined implementation plan, incorporating PoC insights and necessary adjustments for full-scale deployment.
  • Upskilling staff in AI literacy and digital skills through targeted knowledge transfer sessions, ensuring effective platform adoption and governance.

More about the project

You can read the documents related to the project here: