The DOI matches its Crossref record; 10 of the 10 references it cites were found in a registry or an index; 2 of the 15 sources its claims cite were checked for support.
Rethink government with AI
Overview
CoChat Summary
TL;DR
This Nature Comment argues that governments have been slow to adopt artificial intelligence and that data-driven public services could be more responsive, efficient and fair. Drawing on examples of IT failures and one police facial-recognition trial, the authors outline priorities and challenges for embedding AI in government.
Key findings
- Governments struggle with even simpler technologies: HealthCare.gov crashed in 2013; the UK's Universal Credit wasted £837 million (US$1.1 billion) on a digital component that was swiftly decommissioned; Canada's Phoenix pay system has remunerated 62% of employees incorrectly each fiscal year since 2016; and Australia's My Health Record saw more than 2.5 million people opt out by the end of January (roughly 1 in 10 eligible).
- AI could enable government in three ways: personalized services (e.g., Queensland, Australia; New Zealand's SmartStart app), more accurate forecasting (UK local authorities; a Las Vegas–University of Rochester system estimated to prevent more than 9,000 food-poisoning cases and almost 560 hospitalizations per year), and simulation of complex systems (Bank of England housing-market modelling; US disaster assessment; a Mexico agent-computing model for the UN Sustainable Development Goals).
- The London Metropolitan Police's 2017 facial-recognition trial at a carnival flagged 35 "matches"; human reviewers ruled out 30, police stopped the remaining five, and just one was the expected person — who was no longer wanted. The authors use this to illustrate five challenges: low accuracy, high stakes and trust erosion, transparency, appropriateness of individual-level predictions, and untested racial bias.
- The authors argue that data-intensive technologies could lay bare long-standing biases (such as in judicial decision-making) and that a data-driven government "might actually be more fair, transparent and responsive than the human face of officialdom."
Methods
- This is a Nature Comment/opinion piece, not an empirical study; no formal design, dataset or analysis is reported.
- The priorities and challenges presented are "drawn from our experience of working with policymakers at The Alan Turing Institute in London," supported by cited examples and references.
Limitations
- Not stated in the paper (it is a commentary and reports no study-specific limitations).
- Evident from the text: the argument rests on illustrative examples rather than systematic evidence, and the proposed benefits of AI in government are asserted rather than evaluated.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
2 Valid · 2 Not valid · 12 Not checked · 13 Own results and interpretations
Every source the paper cites for this claim was checked and supports it.
In the paper
“The health department of Las Vegas, Nevada, working with the University of Rochester in New York, has used socialmedia data and machine learning to identify restaurants associated with food poisoning.”
What the sources say
- Supportedcites [2] Sadilek, A. et al. AI Magazine 38(1), 37–48 (2017). source
The Abstract (p. 1) states that nEmesis applies machine learning to Twitter (social-media) data to automatically detect venues likely to pose a public health hazard, demonstrated in the Las Vegas metropolitan area 'in collaboration with Nevada's health department', and that the Southern Nevada Health District conducted the three-month controlled experiment flagging the highest-risk venues for inspection. The University of Rochester involvement is documented in the author affiliations (References/author note, p. 4): Henry Kautz, Jack Teitel and Vincent Silenzio are at the University of Rochester, while Lauren DiPrete is a senior environmental health specialist at SNHD. Figure 3 and the Results and Discussion section confirm venues are ranked by patrons' later self-reported foodborne-illness symptoms, i.e. restaurants associated with food poisoning. (LLM, conf=0.90)
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“One such is the systematic bias shown in judicial decision-making over many decades before AI was on the scene 9,10 .”
What the sources say
- Not checked yetcites [9] Spohn, C. C. Crim. Justice 3, 427–501 (2000).
- Supportedcites [10] Spohn, C., Gruhl, J. & Welch, S. Law Soc. Rev. 16, 71–88 (1981) sourceExperimental — accuracy not yet measured
The cited paper’s title identifies race and sentencing as its subject, but its full text is unavailable in Semantic Scholar, so its findings could not be checked directly. A snippet from “Beyond Black and White: An Examination of Afrocentric Facial Features and Sex in Criminal Sentencing” (score 0.59) cites Spohn, Gruhl, and Welch as finding longer sentences for racial minorities, supporting the claim’s point that judicial bias was studied before AI. That secondary source dates the study to 1981 rather than the supplied 1982; the cited paper alone does not establish the broader ‘over many decades’ wording. (LLM, conf=0.72)
boosted- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
In the paper
“Some UK local authorities are experimenting with the use of analytics to anticipate future needs in areas such as homelessness, emergency services and social services 1 .”
What the sources say
- Not checked yetcites [1] Bright, J., Ganesh, B., Seidelin, C. & Vogl, T. Data Science for Local Government (Oxford Internet Institute, 2019).
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
In the paper
“And advisers in Mexico are using an agent computing model 5 to identify what the federal government needs to prioritize to reach the United Nations Sustainable Development Goals.”
What the sources say
- Not checked yetcites [5] Castañeda, G., Chávez-Juárez, F. & Guerrero, O. A. J. Econ. Behav. Organ. 154, 335–361 (2018). source
plainSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
In the paper
“For example, the Bank of England is modelling the UK housing market and simulating the effects of policy measures aimed at mitigating financial risk.”
What the sources say
- Not checked yetcites [4] Axtell, R. Handb. Comput. Econ. 4, 157–213 (2018). source
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
11 more claims
Create a free account to see all 29 claims and to verify the 13 references still open.
- It may state two separate claims as one. 5
- Probably not a substantial claim. 7
- The quote only partly states this claim. 1
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?
The cited source says what the claim attributes to it.
In the citing paper
“Machine-learning techniques are increasingly used to support or automate decision-making processes in government (Margetts and Dorobantu 2019;Vogl et al. 2020).”
From Legitimacy of Algorithmic Decision-Making: Six Threats and the Need for a Calibrated Institutional Response (2022) · cites this paper as Margetts and Dorobantu 2019
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“there are many public organizations where productive applications remain rare (Mikalef et al. 2021; Oxford Insights 2020; Margetts and Dorobantu 2019; Wirtz and Müller 2019).”
From Exploring artificial intelligence adoption in public organizations: a comparative case study (2024) · cites this paper as Margetts and Dorobantu 2019
In the citing paper
“AI may also be used by governments to generate more accurate forecasts and to simulate complex systems that allow experimentation with various policy options (Margetts & Dorobantu, 2019).”
From Implications of the use of artificial intelligence in public governance: A systematic literature review and a research agenda (2021) · cites this paper as Margetts & Dorobantu, 2019
Citations
Cites
The Effect of Race on Sentencing: A Re-Examination of An Unsettled Question
Spohn, C., Gruhl, J., & Welch, S. (1982) · Law & Society Review · doi:10.2307/3053550 · Cited by 186
Cuccaro-Alamin, S., Foust, R., Vaithianathan, R., & Putnam-Hornstein, E. (2017) · Children and Youth Services Review · doi:10.1016/j.childyouth.2017.06.027 · Cited by 145
Shapiro, A. (2017) · Nature · doi:10.1038/541458a · Cited by 118
Information Technology in Government: Britain and America
Margetts, H. Z. (1998) · Cited by 98
Self-citationInformation Technology in Government
Margetts, H. Z. (1988) · Taylor & Francis eBooks · doi:10.4324/9780203267127 · Cited by 58
Self-citationDeploying nEmesis: Preventing Foodborne Illness by Data Mining Social Media
Sadilek, A., Kautz, H., DiPrete, L., Labus, B., Portman, E., Teitel, J., & Silenzio, V. (2017) · AI Magazine · doi:10.1609/aimag.v38i1.2711 · Cited by 56
Castañeda, G., Chávez-Juárez, F., & Guerrero, O. A. (2018) · Journal of Economic Behavior & Organization · doi:10.1016/j.jebo.2018.07.017 · Cited by 40
Data Science for Local GovernmentPreprint
Bright, J., Ganesh, B., Seidelin, C., & Vogl, T. M. (2019) · SSRN Electronic Journal · doi:10.2139/ssrn.3370217 · Cited by 28
Axtell, R. (2018) · Handbook of Computational Economics · doi:10.1016/bs.hescom.2018.05.001 · Cited by 13
Handing NHS data to the Home Office
Gulland, A. (2017) · BMJ · doi:10.1136/bmj.j911 · Cited by 9
Research works printed but not linked: 1
- Spohn, C. C. Crim. Justice 3, 427–501 (2000).
Cited by
Zuiderwijk, A., Chen, Y.-C., & Salem, F. (2021) · Government Information Quarterly · doi:10.1016/j.giq.2021.101577 · Cited by 795
Exploring artificial intelligence adoption in public organizations: a comparative case study
Neumann, O., Guirguis, K., & Steiner, R. (2024) · Public Management Review · doi:10.1080/14719037.2022.2048685 · Cited by 451
“…(ii) the risk of potential erroneous use of AI (e.g. security risks, privacy concerns), (iii) the need to guarantee transparency in the context of AI, (iv) moral dilemmas such as when to use AI, and (v) ethical considerations, (e.g. non-discrimination of citizens) (Margetts and Dorobantu 2019).”
BackgroundInfluentialvan Noordt, C., & Misuraca, G. (2022) · Government Information Quarterly · doi:10.1016/j.giq.2022.101714 · Cited by 356
Digital government transformation in turbulent times: Responses, challenges, and future direction
Eom, S.-J., & Lee, J. (2022) · Government Information Quarterly · doi:10.1016/j.giq.2022.101690 · Cited by 204
Research on international business: The new realities
Ghauri, P., Strange, R., & Cooke, F. L. (2021) · International Business Review · doi:10.1016/j.ibusrev.2021.101794 · Cited by 292
Metadata
- Authors
- Margetts, H., & Dorobantu, C.
- Affiliations
- University of Oxford (GB)
- Corresponding author
- Margetts, Helen
- Venue
- Nature
- Published
- 2019-04-09
- Volume
- 568
- Issue
- 7751
- Pages
- 163-165
- ISSN
- 0028-0836 (ISSN-L)
- 1476-4687
- OpenAlex
- W2937768668
- Semantic Scholar
- 8497f190a562bb141a9cf893bc1db442dadb0cba
- Cited by
- 238
- References
- 10
- Citations by year
- 2026: 60 · 2025: 48 · 2024: 37 · 2023: 31 · 2022: 29 · 2021: 19 · 2020: 10 · 2019: 4
- FWCI
- 65.55
- Type
- journal article
- Language
- en
- Open access
- closed
- Licence
- www.springernature.com/gp/researchers/text-and-data-mining (tdm)
- www.springernature.com/gp/researchers/text-and-data-mining (vor)
- Locations
- Nature (publishedVersion)
- PubMed (publishedVersion)
- London School of Economics and Political Science Research Online (London School of Economics and Political Science) (acceptedVersion)
- RePEc: Research Papers in Economics (submittedVersion)
- Topics
- Social Sciences › Political Science and International Relations › E-Government and Public Services
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Social Sciences › Urban Studies › Smart Cities and Technologies
- Keywords
- artificial intelligence, government, public services, data-driven policy, data governance, public service delivery, administrative efficiency
- MeSH terms
- Facial Recognition, Machine Learning, Big Data, Administrative Personnel, Artificial Intelligence, Computer Simulation, Foodborne Diseases, Forecasting, Ill-Housed Persons, Housing, Humans, Minority Groups, Models, Theoretical, Policy Making, Prejudice, Social Work, Police, Government Regulation, Federal Government, Social Media, Inventions
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.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-09-29, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-09-29 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (qwen3.8-omni-flash and gpt-6-sol).
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
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Rethink government with AI. https://app.cochat.ai/papers/10.1038/d41586-019-01099-5
- The paper
- Margetts, H., & Dorobantu, C. (2019). Rethink government with AI. Nature. https://doi.org/10.1038/d41586-019-01099-5
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.