Q&A: Generative AI technology: a new era in tax management
December 2024 | SPECIAL REPORT: CORPORATE TAX
Financier Worldwide Magazine
December 2024 Issue
FW discusses the impact of generative AI technology on tax management with Romain Tiffon at ATOZ.
FW: Could you provide an overview of how generative artificial intelligence (GenAI) is reshaping tax processes? How open are corporate tax departments to the adoption of GenAI tools?
Tiffon: Generative artificial intelligence (GenAI) will reshape all steps of the tax process. First, we believe that that it will massively influence strategic behaviour by allowing tax professionals to model various scenarios based on a larger set of historical data, while factoring potential changes in tax law to which a probability can be assigned, alongside a thorough analysis of relevant current legislation. Second, GenAI will provide various levels of comfort to tax professionals in many ways. On the deal execution side, GenAI has the power to facilitate the contemporaneous preparation of tax-related documentation, such as tax summaries, transfer pricing and audit reports, or compliance documents. On the compliance side, GenAI has the capacity to extract data, classifying and categorising it in a way that it can populate tax forms, thereby leading to more accurate and faster reporting. This is even more pertinent to companies operating globally that may consequently have to utilise a single data point for several data outputs in multiple jurisdictions. The final point is a more general one. GenAI has the capacity to reduce the impact of labour-intensive, low added value tasks by permitting tax professionals to automate these tasks while focusing more time on high added value tasks, which will in turn contribute to driving innovation in the mid to long term. The adoption of GenAI across corporate departments is quite diverse and nuanced, as it very much depends on the size of the organisation, the industry, and to a greater extent, its technological infrastructure. Clearly, there is increased awareness across the board, but one fundamental hurdle relates to the integration of AI or lack thereof, with existing enterprise resource planning systems, software and other legacy technologies. This is exacerbated by legitimate concerns over data confidentiality and integrity. In addition, there is a conflict between GenAI’s ability to reduce mid to long term costs on the one hand, and the need to weigh in the upfront costs associated with its implementation on the other.
FW: What benefits can be gained by integrating GenAI technologies into tax workflows? In what ways can they transform corporate tax operations?
Tiffon: Integrating GenAI into tax workflows can offer substantial benefits by streamlining operations, enhancing accuracy, improving compliance and enabling more strategic decision making. This technology has the potential to transform corporate tax operations in a variety of ways. It has the power to transform manual input tasks into fully automated ones, notably when it comes to extracting data from various sources, such as invoices, tax forms, financial statements and contracts, and processing that data and populating tax returns and other tax-related documents. As a result, it will reduce the amount of time spent on routine tasks by tax teams and increase their focus on more value-adding activities, thereby improving overall productivity. By the same token, it also reduces the risk of errors associated with manual input. GenAI may further enhance tax planning and strategy development. While this exercise has always been done relying on existing data, GenAI may bring the exercise to a different level by performing a similar analysis, which is likely to be far more accurate due to the significantly greater data set that the GenAI can analyse in a short period of time. This data analysis may also serve to identify tax savings opportunities dynamically, which offers a fantastic competitive edge and will likely create a gap between companies that use this technology and those that do not. From a defensive perspective, given the power of GenAI and its technical association with statistical probability, one could also utilise the technology to flag irregularities or predict areas of potential risk by analysing historical data and identifying patterns that have previously led to audits or compliance issues. This would fundamentally change the way tax teams operate by allowing them to proactively address risks and take countermeasures before the fact. GenAI as applied to the tax industry is an obvious demonstration of what we have witnessed in recent years, namely that businesses will continue to adopt a data-driven decision process. What this means in practice is that data is key, but its effectiveness will be highly influenced by its quality, and therefore its accuracy and structure. GenAI will likely also foster interconnectivity, both as a repository and at user levels. At the repository level, the current issue that companies face is that they typically use siloed applications that do not necessarily communicate altogether, which therefore limits the outreach of the GenAI and may even defeat the very purpose of adopting a GenAI tool. For GenAI to fully reach its potential, there will need to be a change of infrastructure where companies build a data set that is unified and structured. In other words, data would have to be stored in a vault which will store every single data point without duplication, and thus act as the single source of truth. In doing so, a company will also nurture collaboration among its employees, because they will work on the same data set and produce data output that will feed into the repository accessible by other employees.
FW: To what extent can GenAI help tax professionals identify potential risks and compliance concerns, thereby improving accuracy and adherence to regulatory mandates?
Tiffon: GenAI has the potential to significantly enhance tax professionals’ ability to identify potential risks and compliance concerns, as it has the capacity to analyse large data sets and to identify patterns. From that standpoint, GenAI should lead to improved accuracy and adherence to regulatory mandates. In parallel, this may lead to identifying potential risks which could result in tax audits, suggesting remedies and consequently minimising risk exposure and associated penalties. We have seen a surge in tax liability insurance in recent years, but here GenAI could act as the first tranche of an insurance policy when it comes to tax risk management. From a compliance perspective, by leveraging GenAI companies can ensure they file their tax returns on time and in line with applicable legislation. While this is a compliance issue, there is greater pressure on taxpayers to file on time as tax authorities are generally hunting for fiscal revenues. This is on the back of a growing need for companies to have all their data in a single consolidated environment which must, among other things, provide a dynamic dashboard to highlight their compliance completion rates. One additional factor is that they will be able to dynamically adapt to legislative changes. By providing early warning signs of compliance issues, offering real-time updates on regulatory changes and streamlining workflows, GenAI is likely to transform tax operations from a reactive, compliance-driven function into a proactive, strategic one.
“Tax professionals should start by thinking through how GenAI can support their daily tasks, with a view to fully embracing its transformative capabilities.”
FW: What are some of the key challenges of GenAI implementation that corporate tax departments need to overcome?
Tiffon: In the 12 months that followed ChatGPT’s launch in November 2022, everyone spoke about implementing GenAI as if it would be an easy transition. The reality is in fact much more complex, with companies facing significant hurdles to adoption. The first challenge that seems to have been completely overseen relates to data quality and availability. The initial working premise was that GenAI was so agile that it could work on entirely unstructured data sets. In addition to a lack of structure, the data that large corporate groups tend to host on different systems is highly fragmented and not necessarily available in compatible formats, which is a major obstacle to the benefits that GenAI can bring. In this regard, companies should host their data in a structured manner where it becomes the single source of truth on which GenAI will work. Another challenge is organisations’ cultural readiness. Tax professionals, whether external or in-house, tend to be rather conservative in their approach to new technologies, and some are concerned that GenAI will lead to job displacement. Our position is that GenAI should actually support the work of tax professionals by enhancing their skills rather than replacing them. That said, this requires a proper context setting and will clearly impact the training that tax professionals will receive in the future, as they will need to be sufficiently skilled to intelligently prompt large language models but also to challenge the output that is being given to them. This is also likely to reshuffle the pricing models that tax professionals, and to a greater extent lawyers, offer to their clients, and one could imagine that the hourly rate model will transition into an output-based model. A final key challenge pertains to data integrity and confidentiality. Data, and even more so tax data, is highly sensitive. Tax professionals are keen to ensure that the data on which GenAI works remains in a closed, controlled environment and is not used in any way by the model for training purposes. We believe that GenAI tools have already incorporated that fundamental principle by proposing on-premise tools that only work on a defined data set without sharing it with the outside world, while offering training that is grounded on a large data set that is not confidential.
FW: With tax regulations, codes and reporting obligations constantly evolving, how can tax professionals best position themselves to leverage GenAI technologies to deliver faster, more accurate information to senior decision makers at their company?
Tiffon: Tax professionals should rollout their use of GenAI in phases to produce the best leverage for different goals. The first reason for this is to build a more granular understanding of GenAI and its capabilities internally. This is twofold in that it will allow tax professionals to properly formulate requests made to GenAI but also to accurately interpret – and possibly challenge – AI-generated output. This increased understanding and knowledge will also foster management buy-in. The second reason for a phased rollout is that it allows the company to concentrate initially on tasks that are labour intensive but low added value. This then redirects the tax function toward higher added value work and enhances strategic decision-making processes and communication with senior management. This first phase of onboarding should focus on creating a well structured data architecture that will underpin GenAI capabilities for higher-value tasks. The use of small pilot projects prior to going full-scale facilitates a well thought through and well designed GenAI strategy because, while there is a wide variety of tools, having a refined AI strategy will also depend on empirical testing unique to the organisation. At the end of the day, there is no ‘one size fits all’ approach when it comes to positioning. It requires tax professionals to have a thorough understanding of their internal organisation structure and process, and to ensure their data is adequately structured and connected to legislative databases, so that their GenAI setup works on relevant and pertinent data.
FW: What essential advice would you offer to tax professionals on tracking the latest developments in GenAI and assessing the potential opportunities to improve their organisation’s tax planning, compliance and reporting processes?
Tiffon: GenAI is changing daily, so it may be hard for tax professionals to keep pace and stay competitive. The fundamental principle that we would suggest tax professionals follow is to define a clear strategy for implementing GenAI, as this has to be tailored to the company’s needs and organisational structure. Use cases are more or less unlimited, so there is a risk that GenAI may become overwhelming and that onboarding may not be adapted to actual needs. The assessment phase should be structured to outline key needs, current challenges and how GenAI may solve or address them, weighing this against associated costs. This should include how AI can improve reporting speed, accuracy and compliance metrics, which are items senior management consider when deciding how to proceed and what budget to allocate for digital transition. This assessment phase should involve data scientists, as they bring tech insights on a vast array of AI trends and can help tax professionals understand how GenAI tools could be integrated into tax workflows on a task-specific basis.
FW: Looking ahead, what are your predictions for the evolution of GenAI as it pertains to tax operations? Do you believe tax professionals need to embrace GenAI’s transformative capabilities?
Tiffon: We believe GenAI will be truly transformative for tax operations because it allows tax professionals to automate labour intensive, low added value work and allocate resources to more value added strategic work, such as tax planning and advisory, and risk management. GenAI will also help tax professionals design different tax strategies by allowing them to run different models adjusted by current and prospective legislative instruments, while embedding a probability factor. This means that managing tax affairs will become more proactive and forward looking, and thus a strategic item to be factored into asset performance and investment thesis. From a more defensive standpoint, this may also help tax professionals anticipate tax liabilities, which for now is priced only by tax liability insurers. In addition, tax authorities are becoming increasingly digitalised and adopting AI tools to conduct tax audits and review tax filings. This is a significant incentive for taxpayers to be suitably equipped to provide the evidence necessary to match the sophistication of AI-powered audits. Finally, organisations that adopt GenAI tools for their general tax operations, when intelligently rolled out, are likely to gain a competitive edge by optimising their tax positions, avoiding penalties and making data-driven decisions faster than their peers. Tax professionals should start by thinking through how GenAI can support their daily tasks, with a view to fully embracing its transformative capabilities. We are convinced that tax professionals across the board will see improved efficiency, accuracy and strategic insight, with the added benefit of positioning themselves as key contributors to their organisation’s growth.
Romain Tiffon is a partner in the international & corporate tax department at ATOZ. A tax professional since 2006, he has experience in structuring pan-European alternative investment funds across all asset classes, as well as coordinating tax structuring advice and implementation for a wide range of institutional investors. He also has extensive experience in structured finance, M&A transactions and sovereign wealth funds. He can be contacted on +352 26 940 245 or by email: romain.tiffon@atoz.lu.
© Financier Worldwide