What are you looking for?
Back to Outputs
Working Group Recommendation July 23, 2024

Guidance for Informed Consent in the context of Artificial Intelligence and Data Visitation

  • Output Type: Working Group Recommendation
  • Output Status: Endorsed
  • Review Period End: 2024-08-23
  • DOI: 10.15497/RDA00121
  • Standards: TS4 Practical Policies Recommendations
  • Primary Domain: Domain Agnostic
  • RDA Pathways: Other
  • Group Technology Focus: Policy-Related
  • Stakeholders: Funders & Policy makers, Industry & Private Sector, Research Performing Organisations, Researchers & Scientists
  • Sustainable Development Goals: Industry, Innovation and Infrastructure, Partnerships to achieve the Goal, Peace and Justice Strong Institutions, Quality Education, Reduced Inequality
  • Language: English

Abstract

The development and use of artificial intelligence (AI) require access to large amounts of different types of data (data visitation) for training, validation, and refinement of models. This includes data visitation that can be regulated under data protection law, the so-called personal data, which is directly or indirectly identifiable personal information that is subject to confidentiality and to the rights of the individuals concerned (data subject’s rights). In particular, data subjects should be able to express their will regarding their contribution to research by being involved in the process of granting access to data for the development of AI systems. These recommendations guide how such involvement can be envisaged in different technological environments and for different scientific purposes. The guidance considers the conditions necessary to allow for autonomous choices by data subjects through the practice of informed consent, as well as alternative methods that allow for similarly effective individual involvement and expression of will regarding AI data access for research. Recommendations for action to facilitate the design and establishment of such processes are provided for different stakeholders in data governance, namely international organisations, governments, academia, industry, and civil society.

Impact Statement

The recommendations are intended to be adopted by various stakeholders in AI and data governance to promote responsible practices in the innovation and use of AI systems. The goal is both to raise awareness of the important issues raised by AI concerning individual autonomy and to facilitate the development of practical and ethically sound approaches to protecting individual interests throughout the lifecycle of AI systems.

Explanation of Sustainable Development Goals

As outlined in the Guidance, considering individual rights and means for ensuring autonomy and appropriate expression of will regarding data visitation in scientific research for developing AI systems is a necessary step for establishing appropriate conditions for inclusive and responsible research and innovation. Considering this guidance can ultimately be instrumental for enhancing the processes for visiting personal data held by different data holders in respect of individual fundamental rights for scientific purposes, in particular for AI technological development. Therefore, the Guidance contributes to UN SDG 9 (Industry, Innovation and Infrastructure), 4 (Quality Education), 10 (Reduced Inequalities), 16 (Peace, Justice and Strong Institution) and 17 (Partnerships for the Goal).

Citations

Jacob Retanan, L., Dubruel, N., Chassang, G., Hackett, K., Cambon-Thomsen, A., & RDA Artificial Intelligence Data Visitation Working Group (AIDV WG). (2024). Guidance for Informed Consent in the Context of Artificial Intelligence and Data Visitation (Version 2). Research Data Alliance. https://doi.org/10.15497/RDA00121

Internal reviewers: Ademola Adeyeye, Hussein Sherief, Ugochi Adaku Okengwu, Lars Eklund, Mary Uhlmansiek, Dennis N. Mwighusa, Susanne Uusitalo

Broader Working Group: Artificial Intelligence Data Visitation Working Group (AIDV WG)

Metadata information:

– Keywords: Artificial Intelligence (AI), Data Visitation (DV), Informed Consent, AIDV, Ethics, Law, Policy

– Publication date: 14 May 2024

Licensing information: Creative Commons Attribution 4.0 International (CC BY 4.0)

Files

Name Date Added

Comments

  • Profile Picture

    August 22, 2024 at 1:26 pm

    Reyna Jenkyns says:

    The effort to formulate these recommendations on informed consent is much appreciated, as there is clearly a need to re-consider implications as we encourage data re-use and grapple with AI. I have a few comments:
    - please include a definition of data visitation, even if it is not perfect.-
    - the community consent type is also often used in the case of Indigenous communities.
    - the literature review linked from the appendix is not publicly available, permissions should be set to let anyone view with the link
    - it is not clear to me why these recommendations are geared only to the AI and DV situation - seems that they would also be suited to the wider re-use in open science?
    - the recommendations listed all make sense, but perhaps there could be another row that targets data professionals, so that they consider ways to include consent information as machine-readable metadata, develop tooling for dynamic consent, etc.

    Thanks for providing this summary of the landscape and related recommendations going forward.

    Reply
    • Profile Picture

      October 25, 2024 at 1:26 pm

      Gauthier Chassang says:

      Dear Reyna, We thank you very much for your comments that served to enhance the guidance. We have considered them all in the update of the document (please, see the V2 - October 2024) For transparency purposes, you will find more details below. – Please include a definition of data visitation, even if it is not perfect: Initial definitions (as currently discussed by another sub-group of the AIDV-RDA working group) have been included in the introduction and specifically in footnote 6. – The community consent type is also often used in the case of Indigenous communities: This comment has been taken into account by completely rewriting a paragraph on pages 14-15 of the deliverable. – The literature review linked from the appendix is not publicly available, permissions should be set to let anyone view with the link: The access rights for the attached document have been modified to allow anyone to read it. – It is not clear to me why these recommendations are geared only to the AI and DV situation – seems that they would also be suited to the wider re-use in open science? This comment gave rise to discussion within the group, and while we agree with this perspective, we have decided to maintain the focus on AI/DV to avoid going beyond the scope of the document’s objectives. – The recommendations listed all make sense, but perhaps there could be another row that targets data professionals, so that they consider ways to include consent information as machine-readable metadata, develop tooling for dynamic consent, etc. A line of recommendations specific to data professionals has been added (page 20). We hope that these updates answer your comments. Otherwise, feel free to let us know. Best regards On behalf of the AIDV Sub-Working Group on informed consent

      Reply

You must be logged in or join the group to leave a comment.