Editorial: Interdisciplinary Perspectives on the (Un)fairness of Artificial Intelligence external link
Links
Artificial intelligence
Keyword: Artificial intelligence
Dangerous Criminals and Beautiful Prostitutes? Investigating Harmful Representations in Dutch Language Models external link
Abstract
While language-based AI is becoming increasingly popular, ensuring that these systems are socially responsible is essential. Despite their growing impact, large language models (LLMs), the engines of many language-driven applications, remain largely in the black box. Concerns about LLMs reinforcing harmful representations are shared by academia, industries, and the public. In professional contexts, researchers rely on LLMs for computational tasks such as text classification and contextual prediction, during which the risk of perpetuating biases cannot be overlooked. In a broader society where LLM-powered tools are widely accessible, interacting with biased models can shape public perceptions and behaviors, potentially reinforcing problematic social issues over time. This study investigates harmful representations in LLMs, focusing on ethnicity and gender in the Dutch context. Through template-based sentence construction and model probing, we identified potentially harmful representations using both automated and manual content analysis at the lexical and sentence levels, combining quantitative measurements with qualitative insights. Our findings have important ethical, legal, and political implications, challenging the acceptability of such harmful representations and emphasizing the need for effective mitigation strategies. Warning: This paper contains examples of language that some people may find offensive or upsetting.
Links
Artificial intelligence, language models
RIS
Bibtex
Opinie: AI wordt alleen té slim, als we onszelf dom voordoen external link
Abstract
Wie denkt dat ChatGPT een therapeut kan vervangen, onderschat onze complexe sociale werkelijkheid, betoogt filosoof Marijn Sax.
Links
Artificial intelligence
RIS
Bibtex
AI governance in the spotlight: an empirical analysis of Dutch political parties’ strategies for the 2023 elections external link
Abstract
AI-based technologies are having an increasing impact on society, which raises the question of how this technology will be addressed politically. Thereby, political actors have a dual role to play: They can provide investment to enhance the development and subsequent adoption of these systems while also bearing the responsibility of safeguarding citizens from harm. Hereby, the degree of politicization of the topic, i.e. if a topic is part of the public and political debate, has an immense influence on the political approach to tackle the issue. The more a topic is politicized, the more urgency political parties experience to develop concrete governance approaches. Yet, existing research has not analyzed party programs in terms of discourse around artificial intelligence and policy recommendations. This study focuses on the Netherlands and explores how Dutch political parties discuss AI in their political programs for the 2023 election. We conducted a manual content analysis of all party manifestos for the 2023 elections. Our analysis shows that most parties do not place a big emphasis on AI. And if so, most of the policy proposals are rather reactive to issues that happened in the past, rather than taking a prospective governance approach.
Links
Artificial intelligence, governance, Politics
RIS
Bibtex
Do AI models dream of dolphins in lake Balaton? external link
Procedural Justice and Judicial AI; Substantiating Explainability Rights with the Values of Contestation external link
Abstract
The advent of opaque assistive AI in courtrooms has raised concerns about the contestability of these systems, and their impact on procedural justice. The right to an explanation under the GDPR and the AI Act could address the inscrutability of judicial AI for litigants. To substantiate this right in the domain of justice, we examine utilitarian, rights-based (including dignitarian and Dworkinian approaches), and relational theories of procedural justice. These theories reveal diverse perspectives on contestation, which can help shape explainability rights in the context of judicial AI. These theories respectively highlight different values of litigant contestation: it has instrumental value in error correction, and intrinsic value in respecting litigants' dignity, either as rational autonomous agents or as socio-relational beings. These insights help us answer three central and practical questions on how the right to an explanation should be operationalized to enable litigant contestation: should explanations be general or specific, to what extent do explanations need to be faithful to the system's actual behavior or merely provide a plausible approximation, and should more interpretable systems be used, even at the cost of accuracy? These questions are not strictly legal or technical in nature, but also rely on normative considerations. The practical operationalization of explainability will therefore differ between different valuations of litigant contestation of judicial AI.
Artificial intelligence, digital justice, Transparency
RIS
Bibtex
Copyright Liability and Generative AI: What’s the Way Forward? download
Abstract
The intersection of copyright liability and generative AI has become one of the most complex and debated issues in the field of copyright law. AI systems have advanced significantly to allow the creation of fantastic new content but they are also capable of producing outputs that evoke, adapt, or recreate content that is protected by copyright law, sparking several infringement proceedings against AI companies, particularly in the US. With this rapid evolution comes the need to re-examine existing legal frameworks and theories. In this contribution, I would like to focus on liability challenges at the output stage of AI content generation and share some insights from Sweden to finally ponder about possible paths forward.
Links
Artificial intelligence, Copyright, Generative AI, liability
RIS
Bibtex
Dun & Bradstreet: A Pyrrhic Victory for the Contestation of AI under the GDPR external link
Abstract
The CJEU’s ruling in Dun & Bradstreet clarifies how the GDPR’s ‘right to an explanation’ should enable individuals to contest AI-based decision-making. It states that explanations need to be understandable while also respecting trade secrets and privacy concerns in a balanced manner. However, the Court excludes the disclosure of in-depth technical information and also introduces a burdensome balancing procedure. These requirements both strengthen and weaken the ability of individuals to independently assess impactful AI systems, leading to a pyrrhic victory for contestation.
Artificial intelligence, GDPR, Privacy
RIS
Bibtex
Copyright Data Improvement for AI Licensing – The Role of Content Moderation and Text and Data Mining Rules download
Artificial intelligence, Content moderation, Copyright, Text and Data Mining (TDM)
RIS
Bibtex
Towards Planet Proof Computing: Law and Policy of Data Centre Sustainability in the European Union download
Abstract
Our society’s growing reliance on digital technologies such as AI incurs an ever-growing ecological footprint. The EU regulation of the data centre sector aims to achieve climate-neutral, energy-efficient and sustainable data centres by no later than 2030. This article unpacks the EU law and policy which aims on improving energy efficiency, recycling equipment and increasing reporting and transparency obligations. In 2025 the Commission will present a report based on information reported by data centre operators and in light of the new evidence review its policy. Further regulation should aim to translate reporting requirements into binding sustainability targets to contain rebound effects of the data centre industry while strengthening the public value orientation of the industry.
Artificial intelligence, digitalisation, EU law