AI and Tax Administrations: A Good Match
Abstract
This article presents developments related to tax administrations and AI by focusing on three specific areas: (i) audit selection; (ii) guidance to taxpayers; and (iii) improvement of internal efficiency. In addition, it analyses certain challenges and risks, and provides a brief overview of tax administrations’ use of AI in different jurisdictions.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Connect allegedly cost GBP 100 million.
“Although Connect allegedly cost GBP 100 million, it is worth that price insofar as it has helped recover more than GBP 3 billion in taxes since its launch date.”
From Towards eXplainable Artificial Intelligence (XAI) in Tax Law: The Need for a Minimum Legal Standard (2022) · cites this paper as [44]
Over 90% of current tax compliance risk enquiries are triggered by information and analyses generated by Connect.
“Enquiries are currently triggered in the vast majority of tax compliance risk cases (over 90%) by information and analyses generated by Connect.”
From Towards eXplainable Artificial Intelligence (XAI) in Tax Law: The Need for a Minimum Legal Standard (2022) · cites this paper as [44]
Connect is worth its cost.
“Although Connect allegedly cost GBP 100 million, it is worth that price insofar as it has helped recover more than GBP 3 billion in taxes since its launch date.”
From Towards eXplainable Artificial Intelligence (XAI) in Tax Law: The Need for a Minimum Legal Standard (2022) · cites this paper as [44]
Connect has recovered over GBP 3 billion in taxes since it was launched.
“Although Connect allegedly cost GBP 100 million, it is worth that price insofar as it has helped recover more than GBP 3 billion in taxes since its launch date.”
From Towards eXplainable Artificial Intelligence (XAI) in Tax Law: The Need for a Minimum Legal Standard (2022) · cites this paper as [44]
Cites
Strong, D. M., Lee, Y. W., & Wang, R. Y. (1997) · Communications of the ACM · doi:10.1145/253769.253804 · Cited by 1,298 · see citing works
Using deep Q-learning to understand the tax evasion behavior of risk-averse firms
Goumagias, N. D., Hristu‐Varsakelis, D., & Assael, Y. (2018) · Expert Systems with Applications · doi:10.1016/j.eswa.2018.01.039 · Cited by 32 · see citing works
Carrero, J. M. C., & Ribeiro, J. S. (2020) · European Taxation · doi:10.59403/12reg13 · Cited by 4 · see citing works
Flagged first, then by citations
Cited by
Sorted by citations
Towards eXplainable Artificial Intelligence (XAI) in Tax Law: The Need for a Minimum Legal Standard
Kuźniacki, B., Almada, M., Tyliński, K., Górski, L., Winogradska, B., & Zeldenrust, R. (2022) · World Tax Journal · doi:10.59403/2yhh9pa · Cited by 31 · see citing works
- Authors
- Zackrisson, M., Bakker, A., & Hagelin, J.
- Venue
- Bulletin for International Taxation
- Publisher
- IBFD
- Published
- 2020-10-01
- Volume
- 74
- Issue
- 10
- ISSN
- 1819-5490 (ISSN-L)
- 2352-9202
- OpenAlex
- W4390287589
- Cited by
- 13
- References
- 3
- Citations by year
- 2026: 2 · 2025: 4 · 2022: 3 · 2021: 2 · 2020: 2
- FWCI
- 1.49
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- Bulletin for international taxation (publishedVersion)
- Topics
- Social Sciences › Economics, Econometrics and Finance › Economics and Econometrics › Taxation and Compliance Studies
- Social Sciences › Political Science and International Relations › E-Government and Public Services
- Physical Sciences › Engineering › Industrial and Manufacturing Engineering › Robotic Process Automation Applications
- Keywords
- artificial intelligence, tax administration, audit selection, taxpayer guidance, administrative efficiency, AI risks
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
