Page 50 - AI for Good-Innovate for Impact

AI for Good-Innovate for Impact

Use Case – 8: Developing an Artificial Intelligence (AI) - Powered

System for Enhancing Transparency and Accountability in Public
Procurement and Project Monitoring in Tanzania

Country: Tanzania

Organization: The Prevention and Combating of Corruption Bureau

Contact person: Dr Michael Mollel, msamwelmollel@ gmail .com, michael.mollel@ pccb .go .tz

8�1� Use case summary table

Domain Corruption prevention and combating with LLMs
Problem to be Lack of transparency and accountability in public procurement and project
Addressed monitoring processes.
Key Aspects of Deployment of an AI-powered system to monitor procurement processes
the Solution and project execution, providing real-time analytics and reporting to
ensure compliance and detect irregularities.

Technology Natural Language Processing (NLP), Large Language Models (LLMs), Text
keywords Classification, Anomaly Detection, Multimodal Analysis, AI, Data Analytics,
Transparency, Anti-corruption.

Data Availability Data from public procurement and project records are free and accessible,
3000 cases from 2010 (private for recent cases, public for cases older than
10 years).

GPU Not available, but using from collaboration
Metadata (Type Text data (tender documents, evaluation reports, corruption cases, laws,
of Data) and regulations)

Pipeline Inference:
1. Text Generation: helping investigator by generation using generative
AI.
2. Classification: corruption/no,
3. Other: many downstream tasks.

Model Training Machine learning models trained on historical procurement data and
and Fine Tuning outcomes to predict and identify patterns indicating potential fraud or
mismanagement. The method involves Unsupervised pre-training on the
dataset, followed by supervised fine-tuning for specific tasks.

Testbeds or Pilot To be determined.
Deployments

34