GenAI use cases for tax technologists

In our recent communications and collaborations with customers, we’ve seen many tax departments starting to assess how they can leverage generative artificial intelligence (GenAI) tools. A recent report from Deloitte, Generative AI: What should tax directors be thinking about, can help with these assessments. 

The report is authored by partners in Deloitte’s London office. That said, nearly all of the insights and guidance here are just as relevant to indirect tax groups in other countries and regions. 

Much of the report focuses on the risks associated with AI (e.g., bias, hallucinations, ethical considerations, data governance issues and more) as well as the risk management and governance practices that help organisations develop trustworthy AI applications. This risk and governance-related guidance dovetails nicely with the four core principles that guide Vertex’s AI-related development work:

1. Aligning AI with the right use cases.
2. Empowering users with control.
3. Prioritising data security and privacy.
4. Leveraging human intelligence for accountability.

When implementing a new AI solution, the co-authors of the Deloitte report encourage indirect tax groups to: 

• Develop a comprehensive risk taxonomy;
• Identify risks (and their potential severity) for each use case;
• Implement technical controls and non-technical risk mitigation approaches;
• Adapt current governance frameworks to address AI; and
• Integrate responsible AI practices into development activities – and into the organisational culture.

The report highlights GenAI use cases in the tax compliance lifecycle, such as efficiency & productivity enhancements, targeted tax risk reviews, robust M&A due diligence, knowledge management improvements, ecosystem integrations and streamlining procurement activities.   While the list of use cases continues to grow, they all rely on data as the fuel for GenAI. This places higher expectations on the understanding and cleanliness of the data to support the use cases that it is fueling.

I found the report’s guidance to tax leaders responsible for selecting GenAI use case decisions particularly helpful: the two most critical elements to consider, according to the co-report, include:

1. The human effort required to complete the task (or process) without GenAI; and
2. The effort required to fact-check, validate and/or otherwise oversee the output from the GenAI application. 

“If a task requires effort to execute but is easy to validate,” the co-authors point out, “it has the hallmarks of a good use case.”

I have a hunch we’ll be seeing more AI and GenAI use cases as the pace of tax technology transformation continues to accelerate.