Using AI to make government decisions

Neoliberalism

Keir Starmer’s intent is to use AI to ”turbocharge growth”[1].  In Neoliberal “growth” thinking, efficiency and cost savings would be seen as benefits. As cost savings through austerity can have significant negative effects on public services such as the NHS[2], AI cost savings as a public benefit remain therefore doubtful. Neoliberalism is the core reason AI is implemented by the UK government on the unproven promise that it would save money and increase growth. Shaped by a more than a decade of austerity[3], the UK continues to prioritise cost cutting over all else, including effective regulated AI implementation[4], with no apparent improvement to public services.

[1] UK Government, ‘Prime Minister sets out blueprint to turbocharge AI’ (2025) ⟨https://www.gov.uk/government/news/prime-minister-sets-out-blueprint-to-turbocharge-ai⟩ accessed 08/03/25

[2] Tania Arrieta, ‘Austerity in the United Kingdom and its legacy: Lessons from the COVID-19 pandemic’ (2022) 33(2) The Economic and Labour Relations Review 238; David Stuckler, Aaron Reeves, and Rachel Loopstra, ‘Austerity and health: the impact in the UK and Europe’ (2017) 27(suppl 4) European Journal of Public Health 18

[3] Zoe Irving, ‘The Legacy of Austerity’ (2021) 20(1) Social Policy and Society 97

[4] Big Brother Watch, ‘Briefing on the Data (Use and Access) Bill for Second Reading in the House of Lords’ (2024) ⟨https://bigbrotherwatch.org.uk/wp-content/uploads/2024/11/Big-Brother-Watchs-Briefing-on-the-Data-Use-and-Access-Bill-2024.pdf⟩ accessed 22/02/25

Existing Threats to Constitutionalism and the Rule of Law

AI has numerous issues that have already been identified as problematic for liberal democracies. There are issues with accountability and transparency[1], corporate power and data exploitation[2], the mixing of state roles with private actors[3], socioeconomic inequality through job losses[4] and bias[5]. In addition, there are privacy challenges[6] and the doubt that machines can replicate discretionary powers[7]. As we currently see in the USA there is a significant threat to democracy, through non-elected private actors, such as the so called ‘PayPal mafia’, having direct involvement in government actions. As Shoshana Zuboff[8] explains, the penetration of the existing rule of law has been consistently orchestrated by technology companies, initiating the rule of tech[9]. Many UK government functions are already operated by the private US company Palantir[10]. It is as yet unclear how UK government plans to protect the general public from continuous erosion of democratic rights, allegations of backroom deals and favouritism, if this is evidence for AI decision making implementation. AI-driven governmental decisions have repeatedly failed – from the UK Visa Streaming Service[11] to the recent eVisa disaster[12], and from the GangMatrix[13] to the predictive racially biased policing described in the recent Amnesty International Report[14].

Consistent failings are known– for example the lack of implementation of effective and unbiased systems; legislative shifts; unproven positive effects of austerity; state and private actor mix; lack of transparency and accountability; lack of effective remedial action;  considering everyone ‘guilty until proven innocent and lack of consent in facial recognition technology. It seems impossible to anticipate that there will be much benefit to the general public from AI decision making in governmental processes.

[1] Joe Tomlinson, Justice in the Digital State: Assessing the Next Revolution in Administrative Justice (Policy Press May 2019)

[2] Shoshana Zuboff, The age of surveillance capitalism (PublicAffairs 01/19)

[3] Giovanni De Gregorio, Digital constitutionalism in Europe (Cambridge Studies in European Law and Policy, Cambridge University Press May 2022); Hans-W Micklitz and others (eds), Constitutional challenges in the algorithmic society (Cambridge University Press December 2021); Benjamin Moron-Puech, Jeremy Cornaire, and Harrison Colins, ‘Conceptual and legal challenges to the public order of states

[4] Philip Cross, ARTIFICIAL INTELLIGENCE: A threat to middle-class, white-collar jobs? (Techspace rep, Macdonald-Laurier Institute 2023) ⟨http://www.jstor.org/stable/resrep53114.6⟩ accessed 07/03/25

[5] Jennifer Cobbe, ‘Legal Singularity and the Reflexivity of Law’ in Is Law Computable? (Hart Publishing 2020)

[6] Moron-Puech, Cornaire, and Colins (n 7); Zuboff (n 6)

[7] Christopher Markou and Lily Hands, ‘Capacitas Ex Machina: Are Computerised Assessments of Mental Capacity a ’Red Line’ or Benchmark for AI?’ in Christopher Markou and Simon Deakin (eds), Is Law Computable? Critical Perspectives on Law and Artificial Intelligence (Hart Publishing 2020)

[8] Zuboff (n 6)

[9] Giovanni De Gregorio, Digital Constitutionalism in Europe: Reframing Rights and Powers in the Algorithmic Society (Cambridge University Press May 2022)

[10] Rhiannon Mihranian Osborne, ‘NHS England must cancel its contract with Palantir’ [2024] BMJ q1712; privacyinternationalorg & notechfortyrantsorg, ‘All roads lead to Palantir: A review of how the data analytics company has embedded itself throughout the UK’ (2020) ⟨https://privacyinternational.org/sites/default/files/2021-11/All%20roads%20lead%20to%20Palantir%20with%20Palantir%20response%20v3.pdf⟩ accessed 08/03/25

[11] Electronic Immigration Network, ‘Home Office suspends use of digital streaming tool for visa applications after legal action by JCWI and Foxglove’ (2024) ⟨https://www.ein.org.uk/news/homeoffice-suspends-use-digital-streaming-tool-visa-applications-after-legal-action-jcwi-and⟩ accessed 7 March 2025

[12] The Three Million, ‘Widespread identity document errors on digital status (eVisa) accounts’ (2025) ⟨https://the3million.org.uk/news/2025-03-06/widespread-identity-document-errors-digitalstatus-evisa-accounts⟩ accessed 07/03/25

[13] James A Densley and David C Pyrooz, ‘The Matrix in Context: Taking Stock of Police Gang Databases in London and Beyond’ (2019) 20(1–2) Youth Justice 11

[14] .Amnesty International, ‘AUTOMATED RACISM How police data and algorithms code discrimination into policing’ (2025) ⟨ https://www.amnesty.org.uk/predictive-policing⟩ accessed 22 February 2025

Softening of existing regulations

The EU, through the DSA[1], GDPR (General Data Protection Regulation 2016) and EU AI Act 2023, has made some effort to compensate for the negative effects of the continuous digitalisation of our lives. The UK’s GDPR enacted the UK General Data Protection Regulation (UK GDPR) through the Data Protection Act 2018.  However, this implementation is now in the process of being diluted through the Data (Use and Access) Act 2024, especially through Henry VIII powers, softening of regulation regarding automated decision making and weakening of the Information Commissioners Office[2].

This legislative shift threatens UK constitutionalism by undermining transparency, accountability, and checks on Executive power. Such developments cast doubt on the commitment of current and future governments, including Keir Starmer’s current Labour government, to uphold robust constitutional safeguards. Digital constitutionalism is therefore probably ultimately ineffective[3].

The history of massive technological changes

In previous technological periods of change, most notably the Industrial Revolution, the damages to the general public were high. Child Labour was prevalent and needed to be restricted by the adoption of formal Acts of Parliament (for example, the Factory Act 1833; Children and Young Persons Act 1933). Moreover, Luddites, fighting for their livelihoods, were executed[4] and, in essence, it could be argued that the seeds of Communism and the authoritarian USSR were laid during this revolution.  Ultimately, society is yet to overcome the effects of the Industrial Revolution – indeed, Putin’s current apparent desire to attack ex-Soviet Union nations in a quest to restore the lost Russian Empire exemplifies this[5].  As past technological shifts triggered societal upheaval, a momentous, hasty AI implementation, that, similar to the Industrial Revolution, abruptly disrupts the labour market, risks similar destabilisation.

[1] Digital Services Act 2022

[2] Big Brother Watch (n 4)

[3] Petros Terzis, ‘Against digital constitutionalism’ (2024) 3(2) European Law Open 336

[4] Jathan Sadowski, The mechanic and the Luddite (University of California Press January

2025)

[5] Chaim Shinar, ‘Vladimir Putin’s Aspiration to Restore the Lost Russian Empire’, European Review, 25 (2017), 642–54

Conclusion

From discussion in this paper, therefore, it can be concluded that the neoliberal cost saving aspect is doubtful, the existing issues in AI are plentiful and the motivation to execute a well-regulated legal framework seems to be softening – the lessons from history have not been learned.  It is therefore unwise to embark on another major technological change that is badly regulated in law and guided by the Kaldor-Hicks Efficiency[1], “trickledown” economics or neoliberalism.

A return to Keynesian Economics, known for its social aspect of Capitalism, might be a better economic philosophy to implement AI into government. Actual benefits of AI lie, for example, in better health care[2], but this would need a social mindset that neoliberalism does not deliver, while the existing regulations were clearly not sufficient enough to install massive law defying monopolies in foreign states that essentially rule the digital world. Especially worrying is the retraction on existing regulations by the UK government that already backtrack on current legal protections. It seems we are experiencing a re-run of the Industrial Revolution – the harmful effects of which linger on to this day.

It is reasonable to conclude, therefore, that it is likely there will be no benefit to the general public through AI enacted governmental decision making under the current legal and economic system. 

[1] John R Hicks, ‘The Foundations of Welfare Economics’ (1939) 49(196) The Economic Journal 696; Richard A Posner, Economic Analysis of Law (9th, Wolters Kluwer Law & Business 2014)

[2] Junaid Bajwa and others, ‘Artificial intelligence in healthcare: transforming the practice of medicine’ (2021) 8(2) Future Healthcare Journal e188