Bridging the Gap: Why Tax Professionals Struggle to Get Started with Using AI to Support Their Tax Work

In the recent year or so I have been talking to and training tax professionals on artificial intelligence and use in a tax context. For many tax professionals, the journey toward integrating AI into their daily work is marked by uncertainty and hesitance.

This article delves into the multifaceted challenges that challenges the adoption of AI in tax departments and offers insight into potential pathways for overcoming these barriers.

Lack of Clear Understanding of AI Capabilities

One of the primary roadblocks for tax professionals is a fundamental gap in understanding what AI can - and cannot - do. Tax professionals are trained to navigate complex regulations and make nuanced decisions often based on historical precedents and interpretative judgment. The promise of AI - to automate data extraction, process large volumes of documents, or even flag potential compliance issues - can be clouded by misconceptions. Without clear, industry-specific education on how artificial intelligence can serve as an augmentation rather than a replacement, many professionals remain cautious, misinterpreting AI as an unfathomable black box rather than a practical tool.

Regulatory and Compliance Concerns

Tax work often involves sensitive personal and/or financial data, making regulatory compliance non-negotiable. The introduction of AI raises pressing questions about data privacy, confidentiality, and legal liability. Tax professionals are understandably cautious; an error scenario amplified by AI could lead to severe financial penalties, reputational damage, or regulatory breaches. Another area where AI-assisted tax work where I have heard concerns on AI creating uncertainty is when asked to interpretate a new VAT or TP regulation. Tax professionals performing these request using ChatGPT or CoPilot do not necessarily understand that 1) these AI tools are not trained on tax legislation like  Harvey, TaxGPT or the like, 2) it matters greatly how you ask your question (the prompt) where you need to give the AI tool the required information to ensure that it understand what you are asking it to do. The prompt must be Clear, Accurate, Relevant and Specific to gain the best result from the AI – a later article will give more insight to a CARS prompt. Even the use of words that might have the same meaning can make a difference in the result – just like in tax when interpreting the tax legislation.

AI Training and Skill Gaps

Successful AI integration in tax functions is not just about installing new software. It’s about transforming processes, workflows, data and, fundamentally, the skill set of the tax professionals. Many tax professionals have had limited exposure to advanced technologies and ongoing professional development. With a steep learning curve ahead, professionals often perceive the shift to AI-powered tools as a diversion from their core competencies. For many it is not an area of interest and something the IT department should be taking care of.

Adopting AI requires dedicated training and a willingness to invest time and resources to build AI solutions to support the daily tax work - a commitment that many companies find daunting given their already demanding schedules and tight deadlines. And for the companies where they do commit to use of AI it is often Finance that gets the initial priority.

Fear of Job Displacement and Critical Errors

I often meet the argument that tax professionals fear the use of AI. AI’s potential to automate aspects of tax work has stirred concerns about job displacement, eroding traditional roles that have long been the backbone of tax firms. Working with technology in the BIG 4 and in inhouse roles I truly believe that is not a true statement. I think that AI tools should be seen as an amplifier of the tax function in performing tasks that are manual and tedious, that take long time to do. For example, creating the monthly Tax Risk report to the CFO in PowerPoint, extracting data and providing the initial analysis, finding errors in the compliance filings etc. But again, we should always consider the four-eye principle for tax work – even when using AI tools to do the initial draft of a presentation, or executive summary, completing the VAT return, or analysing tax numbers.

Challenges with Data Integration and Data Quality

Many tax functions face the intricate challenge of integrating AI solutions with their existing systems, which are often not set up to handle tax data correctly. These systems may include multiple ERP platforms, each with different configurations and chart of accounts. Such variations can lead to inconsistencies, inaccuracies, and incomplete records. Integrating AI into this complex environment requires sophisticated data processing capabilities and extensive data cleansing efforts to ensure that the insights generated are reliable and actionable. Poor data quality can significantly undermine the effectiveness application of AI, leading to erroneous analyses and potentially costly compliance errors. Consequently, many tax professionals, hesitate to fully embrace AI, viewing the data integration process as a daunting and risky endeavour.

The biggest challenge of all – time

Time to learn and work with AI is one of the biggest challenges that I hear from tax professionals. The demanding nature of their role means they are constantly navigating new or changing tax legislation, adhering to evolving compliance requirements or giving support business strategy such as entry to new markets or M&A. This relentless pace leaves little room for the exploration and integration of AI solutions. The urgency to stay current with the latest regulations takes precedence, causing many to view the time investment needed for AI training and implementation as an unattainable luxury. Often it means that the time to learn and play with AI is something that is done in their free time.