Page 51 - AI for Good-Innovate for Impact Final Report 2024

AI for Good-Innovate for Impact

8�2� Use case description

8�2�1 Description

Public procurement and project follow-up are critical processes for sustainable development 8 - PCCB
and are often plagued by corruption, fraud, and lack of transparency. This is primarily true for
Developing countries, including Tanzania. This use case uses artificial intelligence (AI) and
machine learning (ML) technologies to enhance transparency and accountability throughout
the tendering and project implementation stages and develop the Generative AI tool to
assist the Investigator in their daily work. This proactive approach empowers the Prevention
and Combating of Corruption Bureau (PCCB) and other stakeholders to combat corruption
effectively, ensuring transparency and accountability in the tendering process and project
follow-up.

Current solutions, primarily based on traditional legal and audit mechanisms, struggle to
contend with the scale and complexity of corrupt practices. These limitations have motivated
the project, which introduces an AI and ML-based approach to predict and detect patterns
indicative of corruption.

The proposed AI system, a powerful tool, is expected to process procurement data, including
tender documents, evaluations, and corruption cases, to detect irregularities and assign a
corruption probability percentage (Natural language classification task). Offering an anticipatory
tool empowers PCCB and other stakeholders—judiciary and public procurement regulator
authority (PPRA)—to take preventative action against corrupt activities, thereby enhancing
transparency and ensuring compliance throughout the procurement lifecycle. This system is
designed to augment efforts, making combating and preventing corruption more effective
and impactful.

While the AI-based method promises more proactive and efficient response to corruption,
its reliance on high-quality, large-volume data may pose a challenge. Additionally, the
effectiveness of the AI system hinges on continuous training and updates to stay current with
evolving regulations and corrupt practices.

The benefits of this AI approach are manifold, including the potential for real-time corruption
detection, automated document analysis, and improved resource allocation for investigative
efforts. Conversely, the approach may encounter drawbacks, such as substantial initial data
curation, possible biases within AI models, and the requirement for ongoing technical expertise
to manage and update AI systems.

Ultimately, the project seeks to provide a robust, scalable solution to a chronic problem, with
implications for broader adoption across other nations facing similar challenges in public
procurement.

UN Goals:

• SDG 9: Industry, Innovation and Infrastructure,
• SDG 11: Sustainable Cities and Communities,
• SDG 16: Peace and Justice Strong Institutions

35