arXiv subjects

Vassilis-Javed Khan

Publications and source records attributed to Vassilis-Javed Khan.

4 recordsLinked to original sources

arXiv2026

Runtime Governance for AI Agents: Policies on Paths

AI agents -- systems that plan, reason, and act using large language models -- produce non-deterministic, path-dependent behavior that cannot be fully governed at design time, where with governed we mean striking the right balance between as high as possible successful task completion rate and the legal, data-breach, reputational and other costs associated with running agents. We argue that the execution path is the central object for effective runtime governance and formalize compliance policies as deterministic functions mapping agent identity, partial path, proposed next action, and organizational state to a policy violation probability. We show that prompt-level instructions (and "system prompts"), and static access control are special cases of this framework: the former shape the distribution over paths without actually evaluating them; the latter evaluates deterministic policies that ignore the path (i.e., these can only account for a specific subset of all possible paths). In our view, runtime evaluation is the general case, and it is necessary for any path-dependent policy. We develop the formal framework for analyzing AI agent governance, present concrete policy examples (inspired by the AI act), discuss a reference implementation, and identify open problems including risk calibration and the limits of enforced compliance.

cs.AI↗

arXiv2025

UX Challenges in Implementing an Interactive B2B Customer Segmentation Tool

In our effort to implement an interactive customer segmentation tool for a global manufacturing company, we identified user experience (UX) challenges with technical implications. The main challenge relates to domain users' effort, in our case sales experts, to interpret the clusters produced by an unsupervised Machine Learning (ML) algorithm, for creating a customer segmentation. An additional challenge is what sort of interactions should such a tool support to enable meaningful interpretations of the output of clustering models. In this case study, we describe what we learned from implementing an Interactive Machine Learning (IML) prototype to address such UX challenges. We leverage a multi-year real-world dataset and domain experts' feedback from a global manufacturing company to evaluate our tool. We report what we found to be effective and wish to inform designers of IML systems in the context of customer segmentation and other related unsupervised ML tools.

cs.HC↗

arXiv2024

From Explainable to Interactive AI: A Literature Review on Current Trends in Human-AI Interaction

AI systems are increasingly being adopted across various domains and application areas. With this surge, there is a growing research focus and societal concern for actively involving humans in developing, operating, and adopting these systems. Despite this concern, most existing literature on AI and Human-Computer Interaction (HCI) primarily focuses on explaining how AI systems operate and, at times, allowing users to contest AI decisions. Existing studies often overlook more impactful forms of user interaction with AI systems, such as giving users agency beyond contestability and enabling them to adapt and even co-design the AI's internal mechanics. In this survey, we aim to bridge this gap by reviewing the state-of-the-art in Human-Centered AI literature, the domain where AI and HCI studies converge, extending past Explainable and Contestable AI, delving into the Interactive AI and beyond. Our analysis contributes to shaping the trajectory of future Interactive AI design and advocates for a more user-centric approach that provides users with greater agency, fostering not only their understanding of AI's workings but also their active engagement in its development and evolution.

cs.HC↗

arXiv2021

Self-Organizing Teams in Online Work Settings

As the volume and complexity of distributed online work increases, the collaboration among people who have never worked together in the past is becoming increasingly necessary. Recent research has proposed algorithms to maximize the performance of such teams by grouping workers according to a set of predefined decision criteria. This approach micro-manages workers, who have no say in the team formation process. Depriving users of control over who they will work with stifles creativity, causes psychological discomfort and results in less-than-optimal collaboration results. In this work, we propose an alternative model, called Self-Organizing Teams (SOTs), which relies on the crowd of online workers itself to organize into effective teams. Supported but not guided by an algorithm, SOTs are a new human-centered computational structure, which enables participants to control, correct and guide the output of their collaboration as a collective. Experimental results, comparing SOTs to two benchmarks that do not offer user agency over the collaboration, reveal that participants in the SOTs condition produce results of higher quality and report higher teamwork satisfaction. We also find that, similarly to machine learning-based self-organization, human SOTs exhibit emergent collective properties, including the presence of an objective function and the tendency to form more distinct clusters of compatible teammates.

cs.HC↗