The project aims to build a text summarisation solution that includes AI assurance. The AI assurance component is required to produce an accurate summary of the source text. High reliability testing criteria is needed for natural language texts generated by AI, specifically for use in government departments. We plan to embed these criteria in a user-friendly product that will enable users to assess the trustworthiness of AI-generated content before using it in decision-making processes. This will streamline and enhance the integration of AI in various industries and domains.
Our focus will be on developing the product’s functionality to ensure that it can effectively assure the quality of AI-generated text summaries. These summaries will be shorter versions of longer, complex texts that were generated using state-of-the-art AI language models. These models will be used to create text summaries of longer texts – such as legislation documents. We will conduct workshops with real users to design, develop, and test the product, with the goal of ensuring its rapid adoption within organisations.