arXiv subjects

Malvika Sharan

Publications and source records attributed to Malvika Sharan.

3 recordsLinked to original sources

arXiv2025

Enabling Ethical AI: A case study in using Ontological Context for Justified Agentic AI Decisions

In this preprint, we present A collaborative human-AI approach to building an inspectable semantic layer for Agentic AI. AI agents first propose candidate knowledge structures from diverse data sources; domain experts then validate, correct, and extend these structures, with their feedback used to improve subsequent models. Authors show how this process captures tacit institutional knowledge, improves response quality and efficiency, and mitigates institutional amnesia. We argue for a shift from post-hoc explanation to justifiable Agentic AI, where decisions are grounded in explicit, inspectable evidence and reasoning accessible to both experts and non-specialists.

cs.AI↗

arXiv2024

Professionalising Community Management Roles in Interdisciplinary Research Projects

In this article we discuss community management in interdisciplinary research teams, focusing on recognising and professionalising roles referred to here as the Research Community Managers (RCM). Drawing insights and examples from research and data science projects, we discuss how RCM roles address some of the researchâ€™s most pressing challenges, from promoting best practices for open research and reproducibility to engaging diverse stakeholders in community-led research and ensuring fair recognition for their contributions. We offer a Community Maturation Indicator and share examples of projects from The Alan Turing Institute, the UK's national institute for data science and Artificial Intelligence (AI), where institutionally supported RCM roles were established. With the aim to integrate RCM expertise in teams involved in data science and AI research, we provide an RCM Skills and Competencies Framework. We also propose a roadmap for professionalising RCM roles by improving recognition and rewards, potential career paths and organisational support structures. To systematically sustain and progress these roles, we recommend institutional investment in establishing RCM teams that are empowered to prioritise collaboration, transparency and community-based approaches in interdisciplinary projects, such as in data science and AI. As a team, RCMs are well placed to connect disparate teams, initiatives and resources across the organisation, building more resilient research communities that can achieve greater innovation, improved project outcomes and a strongly connected ecosystem, with impacts extending beyond their narrow contexts.

physics.soc-ph↗

arXiv2024

Unicorns Do Not Exist: Employing and Appreciating Community Managers in Open Source

Open-source software is released under an open-source licence, which means the software can be shared, adapted, and reshared without prejudice. In the context of open-source software, community managers manage the communities that contribute to the development and upkeep of open-source tools. Despite playing a crucial role in maintaining open-source software, community managers are often overlooked. In this paper we look at why this happens and the troubling future we are heading towards if this trend continues. Namely if community managers are driven to focus on corporate needs and become conflicted with the communities they are meant to be managing. We suggest methods to overcome this by stressing the need for the specialisation of roles and by advocating for transparent metrics that highlight the real work of the community manager. Following these guidelines can allow this vital role to be treated with the transparency and respect that it deserves, alongside more traditional roles including software developers and engineers.

cs.SE↗