Introduction
Accepting Administrator Power’s challenge to streamline processes and cut red tape, the USAID Tajikistan Mission pioneered the use of Artificial Intelligence to automate the generation of PPR narratives, reclaiming roughly 1,250 annual staff hours and mitigating the need for extended work days to adhere to critical deadlines. To achieve this, the Mission leveraged DevelopMetrics’ advanced technology, known as the Development Evidence Large Learning Model (DEELM), to automate the generation of PPR narratives.
Project Scope & Methodology
The scope of the pilot project was to leverage machine learning technology to draft two PPR narratives. In the initial phase, DEELM underwent extensive training based on policy documents, annual activity reports, narrative guidance, and all USAID documents in the Development Clearing House. The aim was to acquaint the model with the intricate details, context, and content of USAID’s programs in the specified areas. Drawing on the knowledge acquired during the training phase, DEELM autonomously generated machine-authored narratives in line with the narrative instruction provided. Lastly, a thorough evaluation was conducted, comparing the machine-generated narrative to the traditionally prepared narratives. This comprehensive review assessed coherence, accuracy, and alignment with the established reporting guidelines to ensure the quality standards expected in the PPR.