TTMAG 4
In AI We Trust? Inside the New Era of Auditable AI in Tax
Edition 4 of TTMAG explores what it really means to rely on AI in a field where every decision is regulated, documented, and open to scrutiny. As AI moves from pilots to production, the issue asks under which conditions tax professionals, CFOs and authorities can trust algorithmic systems, and where human judgment must stay firmly in the loop. Through in-depth interviews with academics, heads of tax, technology leaders and founders, this edition looks at governance, data quality, auditability, and the changing skill set of the tax profession in an AI-driven environment.
Topics Covered
1
Inside the New Era of Auditable AI in Tax
AI in tax has shifted from speculation to consequence, with tools now shaping how data is organized, filings are examined, and risks are managed. It argues that trust in AI rests on two parallel tracks – robust data, processes and controls on one side, and empowered professionals who know how to question and govern AI outputs on the other.
2
Governance First, Technology Second
In his interview, Prof. Dr. Robert Risse reframes AI as the last step in a much longer journey that starts with people, processes, and data. He explains why tax teams need to translate “taxable sentences” into process logic, how symbolic and sub-symbolic AI can work together, and why standardizing tax law as machine-readable “rules as code” may matter more than standardizing the tools themselves.
3
Beyond Automation: Building Trustworthy AI in Tax Compliance
Brigitte Baumgartner Garcia breaks down how large language models really work and why retrieval-augmented and hybrid approaches are essential in highly regulated environments. She highlights governance-by-design, input quality (“garbage in, garbage out”), and hybrid-RAG architectures as practical ways to move from AI that merely answers to AI that reasons in line with tax rules and internal policies.
4
From Proof of Concept to Proof of Trust in the Tax Function
Keval Hutheesing looks at what it takes to move from impressive prototypes to AI systems that withstand regulatory and internal scrutiny. He unpacks three pillars of trust – security, accuracy, and compliance – and shows how scoped prompts, curated knowledge bases, detailed logging, and citation-based responses can make AI outputs explainable, traceable, and ready for audit.
5
Trust Starts With Data: Foundations for AI in Tax
Anita Richter brings a tax leader’s view on why fragmented systems and unclear ownership are still the biggest barriers to AI. She discusses cloud migration, data governance under the CFO, and the need for closer collaboration between tax and IT, along with the implications of real-time reporting and e-invoicing for both businesses and authorities as AI becomes part of the compliance fabric.
6
Reshaping the AI-Augmented Tax Workplace
In the closing interviews, Vishnu Bagri and Lukas Zörner explore how AI is changing roles, responsibilities, and workflows for both large organizations and SMEs. They describe future tax teams where AI agents handle ingestion, reconciliation and first-pass analysis, while certified professionals act as “accountability nodes” who review, sign off, and translate insights into decisions – with human-centered platforms making complex compliance manageable for businesses of every size.
Frequently Asked Questions
Visit our FAQ page for more answers.
Still have questions?
Our team is here to help you understand how the Belt System can advance your career.