Artificial intelligence (generative AI) adoption plan: use case prioritization

What it is about

Artificial intelligence trends in the financial world and the development of a management framework that addresses the prioritization, development and implementation of AI and generative AI use cases across the different areas of a financial institution: business, finance, risk, technology and human resources.

The session starts from the premise that artificial intelligence has a growing impact on the global economy and that its application in the financial sector produces benefits that show up in greater efficiency and in institutions’ results.

Session summary

The speaker, a partner at a consulting firm, proposes a practical view for adopting artificial intelligence and generative artificial intelligence in a financial institution: not tackling isolated, ungoverned use cases, but first building a framework with strategy and appetite, governance and committees, policies and procedures, standardized risk assessment, training and suitable technology infrastructure. He explains how the European artificial intelligence act classifies systems by risk (unacceptable, high and the rest), why credit scoring models are high risk and require transparency, explainability and absence of bias, and that whoever uses a third-party tool remains responsible for its results.

On use cases, he shows real examples such as automatic data extraction from documents, code translation in technology migrations, assistants that answer questions about policies and reports, analysis of sustainability information, the use of office assistants and the automation of supplier assessments. He proposes prioritizing them with a benefit, cost and risk analysis, and starting with three steps: the framework, two or three pilots of about three months, and a training plan. In the questions he noted that most institutions still lack an integrated view, that artificial intelligence makes work more efficient but must keep a person in the process, and how to control the risks of hallucinations and of agile methodologies.

Topics covered

• Framework for adopting artificial intelligence: strategy, governance and policies
• European artificial intelligence act: unacceptable, high and other risk
• Explainability, bias and responsibility when using third-party tools
• Real use cases in risk, technology, sustainability and suppliers
• How to prioritize use cases: benefit, cost and risk
• First steps: framework, pilots and training plan
• Questions: real progress of institutions, controlling hallucinations and people in the process