Observe, inspect, modify: Three conditions for generative AI governance

Article Communication

ChatGPT and the AI Act

Natali Helberger et al.

Summary: Being a tech regulator is challenging due to the emergence of generative AI systems like ChatGPT. This essay critiques the European AI Act, arguing that its risk-based approach is insufficient for addressing systems like ChatGPT.

INTERNET POLICY REVIEW (2023)

添加到收藏夹

Article Multidisciplinary Sciences

Large language models encode clinical knowledge

Karan Singhal et al.

Summary: This paper introduces a multi-domain benchmark for medical question answering, which evaluates the performance of models in terms of factuality, comprehension, reasoning, possible harm, and bias through human evaluation. In addition, it proposes instruction prompt tuning to align language models to new domains. The experimental results suggest the potential value of model scale and instruction prompt tuning in improving comprehension, knowledge recall, and reasoning abilities. The human evaluations reveal the limitations of current models and emphasize the importance of evaluation frameworks and method development in creating safe and helpful large language models for clinical applications.

NATURE (2023)

添加到收藏夹

Article Law

AI and Global Governance: Modalities, Rationales, Tensions

Michael Veale et al.

Summary: This review examines different forms of global governance of artificial intelligence (AI) and considers factors such as ethical councils, industry governance, contracts and licensing. The importance of interests and ideas driving these factors is emphasized, and the article urges individuals to constantly question who benefits from global governance.

ANNUAL REVIEW OF LAW AND SOCIAL SCIENCE (2023)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Foundation models and the privatization of public knowledge

Fabian Ferrari et al.

Summary: In order to ensure the integrity of knowledge production, it is necessary to provide regulators and researchers with access to the training procedures of foundational models like GPT-4. Foundation models need to be open and accessible, although they are not synonymous.

NATURE MACHINE INTELLIGENCE (2023)

添加到收藏夹

Proceedings Paper Computer Science, Artificial Intelligence

The Gradient of Generative AI Release: Methods and Considerations

Irene Solaiman

Summary: The article presents a framework for assessing access levels to generative AI systems and highlights the need for diverse perspectives and safety controls.

PROCEEDINGS OF THE 6TH ACM CONFERENCE ON FAIRNESS, ACCOUNTABILITY, AND TRANSPARENCY, FACCT 2023 (2023)

添加到收藏夹

Article Communication

Algorithms as regulatory objects

Robert Seyfert

Summary: This paper discusses the impact of algorithms on social life and related regulatory processes. Regulation is aimed at not only preventing certain algorithmic activities, but also co-producing algorithms. The article illustrates this point through a comparison of different European regulations.

INFORMATION COMMUNICATION & SOCIETY (2022)

添加到收藏夹

Article Social Sciences, Interdisciplinary

The great transformer: Examining the role of large language models in the political economy of AI

Dieuwertje Luitse et al.

Summary: LLMs have become increasingly important in the field of artificial intelligence but have raised concerns about potential biases, environmental impact, and future social consequences. The emergence of LLMs is closely intertwined with the business models of major tech companies, further shifting power relations.

BIG DATA & SOCIETY (2021)

添加到收藏夹

Article Communication

Seeing the forest for the trees: Visualizing platformization and its governance

Jose Van Dijck

Summary: This article uses the metaphor of a tree to analyze platformization and suggest ways for European regulatory bodies and policymakers to address power asymmetry and intervene in the platform ecosystem.

NEW MEDIA & SOCIETY (2021)

添加到收藏夹

Editorial Material Social Sciences, Interdisciplinary