AI Use Cases in GIAA

With innovation as a key strategic focus, GIAA invested in the development of custom AI tools.  

• Insights Engine  - launched in 2022, the Insights Engine is a tool that uses AI and advanced analytics to help auditors process large volumes of text from documents very quickly. It enables users to search, filter, and analyze data across hundreds of documents such as audit reports, board minutes, survey responses, and consultation feedback. The tool has been widely adopted within the GIAA, and feedback highlights its ease of use and ability to efficiently and effectively extract key themes and insights across hundreds of reports. Using AI in this way helps GIAA to quickly identify patterns and insights from audit reports that would be impossible to discern manually.
• Risk Engine - launched in 2023, the Risk Engine leverages generative AI to help auditors identify and articulate risks using a cause-event-consequence framework. This tool enhances risk assessment by providing suggestions for potential risks based on inputted business objectives. This enhances focus on key risk areas, ensuring audits are thorough and relevant to the specific organizational context. It has received positive feedback from internal users and attracted external interest.
• Writing Engine – the Writing Engine is a powerful AI tool developed to streamline the audit report generation process. It integrates with the agency’s audit management system to automatically draft reports based on notes from the audit file (or from a Word template). The engine uses GPT-3 AI to convert auditor notes into structured, clear and concise narratives within seconds. The output serves as a useful first draft and starting point, but it is important for the auditor to carefully review and refine the information to ensure its accuracy, tailoring the report to the specific audit context. The tool avoids hallucinations (unreal or inaccurate content generation) because it is instructed to strictly use the provided input only. The Writing Engine has significantly reduced the time spent on drafting reports, cutting up to two days off the process, while ensuring high-quality, concise outputs. Auditors also benefit from features that suggest audit recommendations and create executive summaries.
  Image

  

In its current research, GIAA is also experimenting using AI to predict future risks by leveraging large datasets, including news articles, to identify emerging patterns and potential threats that may not be immediately apparent to human auditors. This approach involves analyzing trends in external data and assessing how these trends might impact government departments based on their business objectives. The idea is to try and anticipate shifts in risk profiles and proactively consider actions that may mitigate those risks. This capability is still in development but has shown promising results in early stages​​.

Tips for Getting Started with AI

Iain shared valuable tips for organizations looking to adopt AI:

• Focus on processes, not the AI: Begin by identifying critical processes that could benefit from improvement. Consider AI as a tool to enhance these processes, rather than seeking to implement AI for its own sake.
• Don’t delay and start small with one or two use cases to test AI: This allows the organization to focus on managing the risks of those specific cases, which will help in controlling AI's impact and gaining familiarity and more comfort with the technology.
• Build or buy: Decide whether to build AI solutions in-house or purchase them, depending on the resources available. GIAA chose to develop AI internally, but many organizations may opt to use existing tools with embedded AI functionalities.
• Keep people at the center and balance automation with human control: It is crucial for people to feel engaged and confident in using AI, which is why some tasks should be intentionally left to human judgment. For GIAA, this approach helped auditors feel in control of the systems they use. Expect different levels of adoption within the organization. Some employees will embrace AI quickly, while others will need more support and encouragement. Providing one-on-one training and team presentations can help promote adoption, alleviate fears, and overcome resistance.