In Pursuit of Socially-Beneficial Recommender Systems Complexities, Challenges and Interdisciplinary Insights


Recommender systems have now become a ubiquitous feature of digital platforms. They influence user exposure to various forms of content, guiding – sometimes even subliminally but effectively dictating – their online experiences. However, as these systems continue gaining traction and prominence, there is a growing need to critically evaluate their societal impact, with a specific emphasis on a perceived need to promote positive social outcomes rather than merely prioritising the user’s platform engagement and usage time. This paper explores the multifaceted challenges of this endeavour, underscoring first and foremost the formidable complexities of defining what constitutes “positive social outcomes” and for which category of stakeholder those positive social outcomes accrue, and then delving into the challenges of embedding ‘appropriate’ norms and values that align with ‘positive social outcomes’ into recommender systems. This requires a holistic view of user experience, an understanding of the socio-ethical and political-economic dimensions of technology and the economics of technology business, and last but not least the active participation of a diverse array of stakeholders.

To start with, a comprehensive elucidation of ‘positive social outcomes’ must be derived. This is not merely heuristic in scope, but is a fundamental building block – and indeed, the sine qua non – of arriving at any meaningful conclusions with respect to social outcome evaluations, and this elucidation must thus be one that acknowledges the multifaceted, often conflicting interests of different stakeholders, which can range from platform users and profit-motivated owners, to broader socio-political entities and groupings. At the same time, it is critical to address the potential dichotomy between commercial interests, such as maximising platform engagement and commercial engagement valorisation on the one hand, and ethical and regulatory imperatives, like fostering information diversity or maintaining user privacy, on the other. Moreover, it necessitates the delineation of what makes norms and values ‘appropriate’ – a concept that is highly-controversial, context-dependent, chronologically- and spatially- dynamic, and thereby incredibly complex and fraught with uncertainty and subjectivity.

This process involves not just ethical theorisation, but also empirical investigation, consultation with stakeholders, as well as continuous iterative refinement. Ultimately, the integration of these norms and values into recommender systems requires rigorous methodologies, spanning fields from machine learning and data science to philosophy, ethics and various branches of social sciences like sociology, economics, anthropology and political science. In this milieu, this paper explores the delineation of prerequisites for the development of recommender systems as an interdisciplinary, collaborative endeavour, and one that harnesses technological innovation, societal wisdom and political bargaining to deliver not only personalised, but also socially-beneficial digital experiences.

As shown in this paper, the process of defining ‘positive social outcomes’ in an objective, authoritative and incontestable way, might be a much bigger challenge than implementing recommender system features that align with such outcomes, when these have been already properly defined, in the recommender systems themselves. Indeed, once such positive social outcomes are determined and accepted, implementing them in a recommender system should be quite straightforward, even though it might be technically complex, demanding or both.


Recommender systems 87%
Recommender systems norms and values 95%
Social welfare 75%
Social outcomes 50%
User engagement 45%

Access Full Paper

To download the full paper, please click on the PDF icon below. Kindly note that the paper is hosted on the NORMALIZE Conference GDrive (with all conference papers accessible here), an external website hosting papers to which we are not affiliated.

Download PDF

Prefer listening?

If you prefer to listen to, instead of reading the text on this page, all you need to do is to put your device sound on, hit the play button below,  sit back, relax and leave everything else to us.

Algorithmic BrAInIn Pursuit of Socially-Beneficial Recommender Systems Complexities, Challenges and Interdisciplinary Insights

Narration brought to you by

Algorithmic Brain

Looking for other papers you think we might have published but are not able to find?