Accountants: Can AI ERODE Your Tax Expertise?

Tax research is all about painstaking searches, a deep understanding of tax law matters, and a quick but correct comprehension of fast-changing regulations.

That changed with AI coming into the picture!

Over the last few quarters, a new baby (albeit a fast-growing, giant one), artificial intelligence (AI), has generated immense interest as a solution to address the challenges that were almost entry barriers into the tax profession that made tax expertise a specialty.

With AI threatening to permeate the core of tax practices, there is a mounting argument that relying on automation at all costs will undermine the expertise professionals need.

This post is a contrarian look at the current AI trends in tax research to evaluate if AI might stealthily become the de facto preference at the cost of critical human insight.

Rethinking the AI revolution in tax research

For many tax and accounting professionals, AI is simply irresistible. It allows:

• Faster searches and data retrieval,

• Near-instant summaries of complex statutes,

• Unprecedented efficiencies, and

• The capability to instantly create and send client communication emails with just one or two clicks.

The narrative is not that complicated.

However, is there a range of unintended consequences that lurk beneath these attractive benefits?

Technology can undertake routine tasks, but tax research requires much more than speed. It's about precision, context, and nuances that AI may not fully grasp. As professionals increasingly rely on AI, the human ability to delve deeper, i.e., what truly creates expert advice instead of generic output, can be sacrificed.

In my book, "AiCCOUNTANTS: Accountants Augmented by AI," I dedicated a (long) chapter to what humans can do that AI can't.

The biggest truth about AI, which is critical for accountants to protect their human territory, is that AI is super-intelligent based on the data it has been trained on.

You, the human tax pro, must connect the dots between data and decisions (that the clients make or must make).

In my book, I also visualized that "industry-specific" tools would be built on authoritative, vetted tax databases, not broad, unreliable sources. Such tools would be transparent, with easily verifiable references to the answers they produce.

Such tools are intended to augment human expertise.

I recently explored the workings and thought processes behind creating Blue J, an AI tax research solution.

It is critically important that it can refine the research focus based on your inquiries, suggest relevant follow-up questions, and provide citations from credible sources. This enables you to delve deeper into foundational information from authorized sources. It is encouraging to see the commitment to developing AI tools responsibly.

With great tools comes great responsibility

AI systems can analyze large databases in seconds, which is beneficial when handling extensive sources of regulations and case law, especially during the tax season's time crunch. AI tools can also summarize long legal texts into a few words of insight, saving practitioners time reading the key points.

However, if such tools are not used responsibly, this race toward efficiency has some setbacks:

• Loss of context: When AI systems condense intricate legal arguments into summarized, simplified portions, they can lead to a human tendency to overlook the subtly crucial facts for accurate, relevant, and sound tax work and advice. Note that it is not the AI tool but human fallacies that can cause such challenges.

• Risk of Error: Automation can introduce "hallucinations," where AI generates plausible-sounding but incorrect conclusions. These errors can lead to flawed strategies, compliance issues, and penalties if not monitored. As an expert tax professional, your intuition will help you determine if there is adequate "explainability." Can you clarify to your clients and regulators what actions you took based on what AI informed you? You must not become a messenger between AI and clients or regulators.

The pitfalls of purely data-driven decision-making

AI tools can make predictions based on historical data and pattern recognition. However, tax law is a dynamic field that is continuously evolving due to political, economic, societal, and global trade trends. Relying solely on historical patterns can cause AI systems to overlook emerging issues or misinterpret new regulatory environments.

As a result, the expected efficiency gains may turn into liabilities, giving a false sense of security while inadvertently neglecting crucial updates. This is why purpose-built, specially-trained, industry-specific AI tools continuously updated through human-supervised training are essential for tax preparation and planning tasks.

When automation outpaces expertise

As a tax professional, you bring intuition and interpretive skills that, quite frankly, AI and its algorithms cannot easily replicate.

Consider these aspects of human judgment:

• Much of the language in tax laws can lack absolute precision. These laws are written with multiple meanings to cover a variety of contexts. As a human tax professional, you must consider ambiguities in light of a client's specific circumstances.

• Tax strategy is not only a technical exercise but also requires ethical judgment. Therefore, balancing regulatory compliance with fairness and moral responsibility is crucial and relies on human capabilities.

• Unlike static models, human experts can utilize various facets of interdisciplinary knowledge and experience to navigate unexpected/unanticipated challenges. Adaptability to regulatory confusion is essential.

New professionals, trained primarily on AI outputs, might find themselves less capable/confident of critical analysis when faced with unprecedented scenarios. Over-reliance on AI can lead to homogenization or commoditization of tax advice, i.e., one that mirrors past patterns without genuinely engaging with the unique challenges of the present.

Don't compete with what AI does best.

• Rather than adopting a "more is better" approach to AI, as a tax professional, you would want to strive for a balanced integration that leverages technology without sacrificing human judgment.

• Don't compete with what AI does best, such as data aggregation, initial document review, routine calculations, super-fast research, etc., while reserving the interpretive work for experienced professionals.

• Implement continuous quality checks to deploy your human expertise to review AI outputs, ensuring that no critical nuance is missed.

• Undertake continuous professional development emphasizing critical thinking and adaptive problem-solving, even as AI becomes more embedded in daily workflows.

How to evaluate AI Tax Research tools

Knowing about the following is critically essential for you to evaluate any tax research AI solutions:

• Transparency about the continuous development methodologies of the AI tool

• Expertise of the developer company to "supervise" the AI tool's training from a tax experience point of view

• Their data sources,

• The recency of updates, and

• The tool's in-built decision-making processes (how does it arrive at what it shows you).

This transparency in design allows tax professionals to understand the technology's limitations better and respond appropriately when necessary.

The future is now: A challenge to responsibly implement AI.

AI will continue to fulfill its potential and promise in tax research, and it will do so increasingly better as time passes.

This approach will offer the best of both worlds: harnessing AI's rapid ability to process information while maintaining the critical, nuanced analysis that only human professionals can provide.

This means that the counterintuitive approach may be the most effective in a profession where precision is critical, and the stakes are high.

By resisting the impulse to automate every facet of tax research, professionals can ensure that the practice of tax law continues to enjoy the trust it has earned over decades.

The future of tax research is not to abandon AI or embrace AI with wanton abandon. It is about enhancing the value of human judgment. In an era of automation, sometimes the most innovative approach is to do less (as a human, but it does not mean don't do what's essential but get it done using responsible technology) and think more.

If you have any thoughts or experiences on balancing AI with tax research expertise, please share your insights in the comment section below.

If you feel your team members will benefit from this post, please share it with them.

Do you disagree with the thoughts in this post? A debate will only make things better for the profession. CLICK HERE to share your thoughts.