AI in tax: opportunity, risk and the enduring role of professional judgement

These themes were explored at a recent joint Tax Research Network and CIOT webinar on AI in tax research, education and the profession, bringing together insights from educators, advisers, policymakers and senior figures in tax administration.

The discussion followed the publication of the updated Professional Conduct in Relation to Taxation (PCRT), accompanied by topical guidance reiterating that responsibility for tax advice always rests with the adviser, regardless of the tools used. With that context, the webinar focused less on compliance mechanics and more on what responsible, credible AI use looks like in practice.

From the tax administration perspective, Ruth Kennedy (Irish Revenue), Anna Kwiatkowska (HMRC) and Anthony Pritchard (Welsh Revenue Authority) each described how AI is already reshaping organisational decision making. A shared theme was that AI’s real impact lies not in automating existing tasks, but in rethinking how work is organised, how risks are assessed and how decisions are supported.

Kennedy emphasised that revenue bodies are complex systems of processes, tools and institutional responsibilities. AI can enhance productivity and effectiveness, but only when embedded within strong governance frameworks and subject to human assurance. Kwiatkowska highlighted HMRC’s longstanding use of AI and advanced analytics in compliance targeting, debt-risk prediction and customer contact, placing these developments within HMRC’s broader transformation strategy.

Pritchard, representing a devolved administration, outlined how the Welsh Revenue Authority is using AI to support compliance and risk assessment while embedding transparency, organisational values and human oversight.

Implications for advisers were explored both practically and conceptually. Ritchie Tout (Azets Partner) focused on AI in tax valuation, stressing that AI outputs must always undergo critical professional review, particularly where valuation uncertainty and potential penalties arise.

Stephen Daly (King’s College London) reflected on AI use in technical tax research, drawing on his experience as Editor of the British Tax Review. He highlighted risks to academic integrity, noting that AI can generate hallucinated sources, placing greater responsibility on reviewers to check bibliographies thoroughly.

From an education and standards perspective, Kelly Sizer (Head of Qualifications Development, CIOT/ATT) discussed how AI affects assessment, skills development and qualification design. She emphasised the need for continual review of professional qualifications to ensure they reflect technological change while preserving judgement, ethics and technical competence. Terry Filer (TRN Deputy Chair) similarly noted that AI literacy is becoming a core professional capability, but must be paired with critical thinking rather than passive reliance on tools.

Taken together, the webinar underscored a consistent message: AI can enhance efficiency, insight and consistency across tax work, but it does not dilute professional responsibility. As AI absorbs more routine analytical tasks, the distinctive value of tax advisers will increasingly lie in interpretation, judgement and ethical reasoning. AI may change how tax work is carried out, but trust in tax advice – and tax systems more broadly – will continue to rest on human judgement.