Artificial intelligence is creating waves in every industry. From streamlining burdensome tasks to analyzing troves of data, AI presents an incredible opportunity to enhance the way each area of a business — including the tax function – runs its operations.

AI’s abrupt rise isn’t as sudden as it seems. The fine tuning of “emerging tech” over the last few years lends itself to AI’s recent proliferation and accessibility. Cloud storage and security, API-first software, and data transformation tools have paved the way for manageable data stewardship conducive to the technology. In other words, what may feel like daunting new terrain is actually a well-traveled landscape cultivated behind closed doors and now available to the public.

But implementing AI isn’t as simple as a snap of the fingers. Possessing an effective data-governance strategy and understanding how to organize critical information can help tax departments — and their organizations — turn AI into a powerful asset. To get the most out of the rapidly advancing technology, one must first lay important groundwork in key areas.

Establishing Data Governance

Data is the building block of AI systems and serves as the starting point for implementing any new technology solution. For AI to use data most effectively, it must be consistently maintained according to high standards. That requires an effective data governance strategy to keep tax data organized, including:

• Master data management
• Data quality controls
• Collaboration with the enterprise

Through various solutions that include holding data in raw, estimated, and as-filed forms, organizations must ensure they have taken appropriate measures to access high-quality data efficiently. For tax departments, reaching that point will mean creating a uniform data structure that AI can reference with consistent accuracy. Enacting governance policies that way from the beginning will help generate effective results.

The need for strong governance policies also speaks to the disparate nature of tax data — which itself does not infer poor data quality. Tax professionals inherently institute checks and balances to produce materially correct results. Creating high-quality data is more often related to whether it is relevant to the question at hand rather than if a balance is correct. At the same time, proper quality benchmarks can yield insights and increase transparency, which will provide confidence in the data.

Creating Master Data Management

In a perfect world, every piece of data would be uniformly structured and digitized, but that’s simply not a realistic starting point for tax functions. AI was made for the real world, which can be chaotic. Instead of striving for perfection, tax departments and their organizations should start with a consistent methodology that structures data to make it work with data that can’t be uniformly structured. By digitizing as much data as possible and carefully applying an agreed methodology, organizations can reduce disruptions caused by new or dissonant data, be it a new business acquisition or ERP system.

In creating master data, tax departments should assess the tools they already have. While many organizations use tax engines for calculation and compliance purposes, technology can also create structured data as a beneficial output. In fact, implementing a tax engine should be seen as a necessary steppingstone to AI. Departments that have already taken that step are on the right path; those that haven’t should consider it an ideal and proven way to bring structure to disparate tax data. The fundamentals of data automation remain unchanged, even if the vehicle getting you from point A to point B has.

Establishing master tax data generates benefits for the organization that include:

• Creating standardization and improved analysis
• Reducing manual reconciliations
• Leveraging the same data for multiple needs and uses
• Improving naming conventions and data storage
• Accelerating the pace of change in support of business growth

Those benefits may look familiar: They’re the same as those sought in implementing any tax technology. The difference is that master data looks to achieve them across the tax function, rather than for only one particular process.

The Importance of Data Quality

AI development for use by tax departments hinges critically on data quality. High-quality data is the cornerstone for training an AI model that can accurately predict, analyze, and process tax-related information. With the complexity and variability of tax regulations and financial data, ensuring the integrity, accuracy, and completeness of data is paramount. Poor data quality can lead to erroneous AI predictions, misclassification of transactions, and, ultimately, a loss of trust in the platform. Several criteria determine data quality:

Enhancing Data Quality for AI Development

Improving data quality involves several key steps that are critical to prepare the data for AI applications:

1. Data Assessment and Profiling

Assessing the current state of data used by tax teams involves profiling the data to understand its source, structure, content, and quality.

2. Data Cleansing

Correcting inaccuracies, filling in missing values, and resolving inconsistencies in the data are crucial steps to verify the data is accurate and reliable.

3. Data Integration

Integrating tax data, which often comes from multiple sources, in a coherent manner helps create a unified view, which is essential for effective AI analysis. This step may involve aligning different data formats and schemas while confirming consistent data definitions.

4. Data Standardization

Standardizing data formats, values, and nomenclatures helps establish consistency across the dataset. This is particularly important for AI models to accurately interpret and learn from historical data.

5. Data Validation

Implementing rules and constraints to validate data sources can prevent quality issues and fold future tax inputs into the established data standards. This includes establishing checks for data range, format, and logic to create data integrity.

6. Continuous Monitoring and Maintenance

Creating high-quality data is not a one-time effort. Continuous monitoring, regular updates, and maintaining data quality infrastructure are essential to adapt to changes in master tax data, tax laws and regulations, and AI model requirements.

7. Documentation and Governance

Documenting the data quality processes, standards, and policies improves transparency and provides a reference for ongoing data management. Establishing a tax data governance framework helps enforce these standards and policies.

Implementing AI Securely

Creating accessible information comes with inherent risks. Handling sensitive data requires sound security practices, which become even more important when the information exists in the cloud. To protect information, organizations need transparent data management protocols and access controls so that data remains available to the people and systems that need it, while staying out of the hands of those that don’t.

That isn’t to say the entire burden falls on tax teams. AI standards must be established, implemented, and governed at the enterprise level. While tax departments play an important role, secure, ethical, and quality adherence to those standards falls on the entire organization. As intimidating as they may seem, the technical nuances and secure integration of AI is a team effort that should be built on open communication between information technology and tax departments.

Effectively Integrating a Tax AI Strategy Within the Enterprise

Tax teams own their AI strategies, but integration with the enterprise’s overall strategy can come with its own challenges. Although tax departments can benefit from AI in numerous ways — identifying patterns in data to improve tax reporting and planning, creating summaries of complex tax concepts, identifying potential areas of tax risk, etc. — organizations might be hesitant to adopt such a transformative technology quickly. Often, they wait for the pieces to fall into place, rather than being proactive.

But that also creates a catch-22: Organizations hesitate to move the needle on AI because they’re waiting for the right data to come together, but that won’t happen unless organizational leaders strategically develop plans to prepare the data for AI use.

It doesn’t have to be that way, though. By taking small steps, claiming quick but meaningful wins, and gradually introducing AI into specific functions, tax departments can set an example that demonstrates the technology’s efficacy. Doing so allows them to learn from early accomplishments, refine best practices for success, and identify broader uses of AI by the organization. Ideally, that allows the organization to holistically view its tax data and different tax technology platforms to interface. For example, most organizations have established dashboards for tax reporting, financial reviewing, or executive communication. Many intelligence platforms have recently added AI capabilities, such as a question-and-answer feature against the dashboard’s data model, suggesting next steps for data transformation, or even recommending dashboard layout based on historical use.

Working With AI Responsibly

Even after successfully integrating AI tools into daily operations, the job isn’t done. Strategic data governance policies, secure access to information in a cloud-based environment, and cohesive systems that work in unison still require human oversight. No matter how impressive it may seem, AI is still a tool, and its reliance on quality data never ends. The need for judgment calls regarding context, applicability, and AI’s impact on the broader organization also never stops. At the end of the day, the C-suite will hold tax leadership accountable for communicating their executive findings, areas of exposure, and tax planning strategies.

AI is unable to reason and can’t always determine the context of information it’s processing, which can result in outputs it deems accurate but are actually off the mark. In that way, AI is no different from basic technology like a calculator: It will provide an answer based on information the user provides. The difference between the two is the complexity of the information they take in. A calculator might provide an inaccurate answer because of a mistyped number; AI can return the wrong results if files contain inaccurate or mislabeled information. In other words, AI remains contingent on knowledgeable human contributions to work effectively.

AI is already providing tangible benefits to tax departments, and its influence will continue growing as the technology evolves. But no matter how powerful AI becomes, the need for quality data and the expertise of skilled tax professionals will remain.

Copyright © 2024 BDO USA, P.C. All rights reserved. www.bdo.com