Pham and Davies (2025): EU AI Act Problem Representations

A set of 50 vocabulary flashcards examining the problem representations, analytical framework, and broader policy effects of the EU AI Act based on Pham and Davies (2025).

Paper's Central Question

An inquiry into what problems are represented as being solved by the EU AI Act, and what broader effects these representations produce.

WPR Framework

An analytical approach standing for 'What's the problem represented to be?', which examines how policy solutions represent and constitute problems.

WPR Approach to Policy Problems

A perspective treating policy as actively helping bring problems into being, rather than simply responding to pre-existing problems.

Problem Representation via Policy Solution

The principle that a proposed solution inherently carries an underlying understanding of what the problem is, presenting it as a claim about the world.

Core WPR Questions

Questions addressing what assumptions a problem representation contains, how it arose, what is silenced, what effects it produces, who enabled or resisted it, and how the analyst's own assumptions should be questioned.

Three Effects of Problem Representations (WPR)

The three distinct types of policy effects distinguished by the WPR framework: discursive effects, subjectification effects, and lived effects.

AI Act Policy Solutions

The specific policy solutions identified in the AI Act, consisting of bans on some applications, regulatory sandboxes, and strict requirements for certain high-risk AI systems.

Risk-Based Approach (AI Act Analytical Focus)

The paper's main analytical focus within the AI Act, centering on the categorization of AI applications into risk categories.

Two Central Problem Representations of AI

The dual representations identified by the authors: that Europe may miss AI's opportunities and benefits, while AI simultaneously may threaten European citizens' fundamental rights.

Opportunity-Focused Framing

A policy representation framing AI as a source of novelty, optimism, excitement, confidence, and potential social and technological benefits.

"Missing Out" Problem

The problem representation stating that without active participation in AI development and use, Europe risks losing benefits and falling behind other nations.

Expected Benefits of AI

The societal and economic outcomes linked to AI in policy discourse, including productivity, economic growth, social progress, improved medical care, and less pollution.

"Race" Framing

A policy discourse constructing a global, zero-sum technological race in which Europe risks losing out if it does not act.

Non-Risky AI Applications

AI applications not categorized as risky under the AI Act, which remain largely unregulated or subject only to voluntary assessment.

AI Act Risk Scale

A regulatory mechanism connecting risk categories and intended use or sector to varying levels of conformity assessment, documentation, and transparency obligations.

Threat-Focused Representation

The central problem representation framing AI as a potential threat to European fundamental rights and values.

Potentially Threatened Rights and Values

The specific rights and values identified as under threat from AI: privacy, human dignity, equality, personal-data protection, and free movement of goods and services.

High-Risk and Unacceptable-Risk Concern

The underlying concern that AI systems may escape human oversight or develop in ways that conflict with European values and fundamental rights.

Prohibited AI Applications

Applications banned in the analyzed proposal, specifically those involving manipulation, social scoring, biometric identification, or real-time facial recognition.

Proper Role of AI Systems

The normative policy standard dictating that AI systems should support human agency and fundamental rights, and not diminish, limit, or misguide human autonomy.

Tension Between AI Visions

The contrast between framing AI as a solution to major challenges versus framing AI itself as a direct source of risks such as misuse, discrimination, and reproduced biases.

Reconciliation Challenge

The policy challenge of reconciling technological leadership and development with protecting rights and values, without unduly constraining innovation.

Risk-Based Approach (Reconciliation Function)

The AI Act's structural mechanism that reconciles competing problem representations by supplying a single regulatory solution aimed at supporting competitiveness while preventing harm to rights and values.

Necessity of Regulation

The policy assertion that without regulation, Europe will be economically disadvantaged by missing AI benefits while also leaving fundamental rights unsafe.

"Trustworthy AI" Discourse

A wider policy discourse in which the AI Act and its risk-based approach are embedded, intended to make AI develop in ways that citizens can trust.

Qualities of Trustworthy AI

The proposed attributes of trustworthy AI—being legal, ethical, robust, and human-centred—notwithstanding that trustworthiness itself is not concretely defined.

Coherence Role of "Trustworthy AI"

The discursive role of 'trustworthy AI' in making two contrasting visions cohere by presenting AI as both an opportunity and a threat that can be managed through ethical development.

"Win-Win" Proposition

The framing wherein respecting fundamental rights is portrayed as both valuable in itself and as a catalyst for public acceptance and competitive advantage for European AI firms.

Dual Purpose of Value-Aligned AI

The dual function of AI systems aligned with European values to protect essential principles while simultaneously increasing uptake, profitability, and global competitiveness.

Synthesis of AI Problem Representations

The article's conclusion that the AI Act's risk-based approach and trustworthy-AI discourse jointly hold together two contrasting representations: missing AI's benefits and suffering infringements of rights and values.

"Making Policy, Making Europe"

The concept that European AI policy not only shapes how AI is understood but also actively helps constitute Europe and Europeanness.

Europe as a Collective

The representation in policy documents of Europe as a community unified by shared values and a common moral compass centred on citizens' rights.

Distinctive European AI Trademark

The presentation of European AI as human-centred, trustworthy AI that safeguards societal values while establishing a distinctive European trademark.

European Exceptionalism in AI

The portrayal of Europe as possessing a morally distinctive, potentially superior approach to AI governance that serves as a model for other parts of the world.

Harmonization Effect

A policy outcome presenting the EU as a unified bloc acting as one, which blurs national differences and limits national regulators' room to manoeuvre.

Geopolitical Authority Reinforcement

The positioning of the EU as a moral leader, granting it authority and agency to promote its regulatory AI approach internationally.

Subjectification Effect (AI Policy)

A specific WPR policy effect identified in the study through which a subject, in this case 'Europe' itself, is constituted in a particular way.

Single European AI Market Creation

The role of risk-based AI regulation in actively helping to constitute a globally competitive common European AI market.

Brussels Effect

The mechanism by which EU laws diffuse internationally to become global norms, positioning Europe as a moral leader in geopolitics.

Exclusion in Global AI Policy

The critical concern raised by the authors regarding whose rights and values may be excluded when economically developed countries dominate international AI-policy debates.

Overlooked Populations in Citizen-Centric AI

Non-citizen groups, such as people on the move and people affected by global injustices connected to AI, who are rendered invisible when policy centres 'European citizens.'

Citizen Conception ("Homo Economicus")

The tendency of policy documents to conflate citizens with consumers, aligning citizenship primarily with market participation rather than broader political or civic agency.

Incomplete Conception of Citizenship

The critique that equating citizenship strictly with participation in markets omits other possible forms of democratic citizenship and technological governance.

Political Nature of Risk Categories

The insight from the WPR approach showing that categorizing and regulating AI applications by anticipated risk reflects political choices, rather than natural or necessary groupings.

Ambiguity of "Trustworthy AI"

The critique highlighting that 'trustworthy AI' lacks a concrete practical meaning and has been characterized by scholars as a buzzword or an empty signifier.

Self-Problematization

A reflexive requirement in WPR for researchers to examine their own assumptions and acknowledge how their analysis may contribute to producing the realities they describe.

Excluded Corpus Actors

Key external entities—including industry, media, national policy, NGOs, and civil society—that shape AI and European identity but fell outside the study's primary corpus.

Contingency of AI Policy

The authors' perspective that current AI policy trajectories are not inevitable, meaning alternative approaches to AI governance and development can be imagined.

Broader Risks of AI Policy Discourse

The systemic risks that EU policy activity may reinforce techno-solutionism, technological determinism, and limited versions of European citizenship.

Overall Conclusion on the EU AI Act

The paper's summary finding that the AI Act's risk categorization and trustworthy-AI discourse reconcile the 'missing out' and fundamental-rights problem representations, while actively producing the EU as a coherent, values-based collective and moral leader.