Katie Winkle

Ines Hipolito, Katie Winkle, Merete Lie

This paper introduces Enactive Artificial Intelligence (eAI) as an intersectional gender-inclusive stance towards AI. AI design is an enacted human sociocultural practice that reflects human culture and values. Unrepresentative AI design could lead to social marginalisation. Section 1, drawing from radical enactivism, outlines embodied cultural practices. In Section 2, explores how intersectional gender intertwines with technoscience as a sociocultural practice. Section 3 focuses on subverting gender norms in the specific case of Robot-Human Interaction in AI. Finally, Section 4 identifies four vectors of ethics: explainability, fairness, transparency, and auditability for adopting an intersectionality-inclusive stance in developing gender-inclusive AI and subverting existing gender norms in robot design.

Katie Seaborn, Madeleine Steeds, Ilaria Torre et al.

The "gender" of intelligent agents, virtual characters, social robots, and other agentic machines has emerged as a fundamental topic in studies of people's interactions with computers. Perceptions of agent gender can help explain user attitudes and behaviours -- from preferences to toxicity to stereotyping -- across a variety of systems and contexts of use. Yet, standards in capturing perceptions of agent gender do not exist. A scoping review was conducted to clarify how agent gender has been operationalized -- labelled, defined, and measured -- as a perceptual variable. One-third of studies manipulated but did not measure agent gender. Norms in operationalizations remain obscure, limiting comprehension of results, congruity in measurement, and comparability for meta-analyses. The dominance of the gender binary model and latent anthropocentrism have placed arbitrary limits on knowledge generation and reified the status quo. We contribute a systematically-developed and theory-driven meta-level framework that offers operational clarity and practical guidance for greater rigour and inclusivity.

Alessio Galatolo, Luca Alberto Rappuoli, Katie Winkle et al.

The recent rise in popularity of large language models (LLMs) has prompted considerable concerns about their moral capabilities. Although considerable effort has been dedicated to aligning LLMs with human moral values, existing benchmarks and evaluations remain largely superficial, typically measuring alignment based on final ethical verdicts rather than explicit moral reasoning. In response, this paper aims to advance the investigation of LLMs' moral capabilities by examining their capacity to function as Artificial Moral Assistants (AMAs), systems envisioned in the philosophical literature to support human moral deliberation. We assert that qualifying as an AMA requires more than what state-of-the-art alignment techniques aim to achieve: not only must AMAs be able to discern ethically problematic situations, they should also be able to actively reason about them, navigating between conflicting values outside of those embedded in the alignment phase. Building on existing philosophical literature, we begin by designing a new formal framework of the specific kind of behaviour an AMA should exhibit, individuating key qualities such as deductive and abductive moral reasoning. Drawing on this theoretical framework, we develop a benchmark to test these qualities and evaluate popular open LLMs against it. Our results reveal considerable variability across models and highlight persistent shortcomings, particularly regarding abductive moral reasoning. Our work connects theoretical philosophy with practical AI evaluation while also emphasising the need for dedicated strategies to explicitly enhance moral reasoning capabilities in LLMs. Code available at https://github.com/alessioGalatolo/AMAeval

Alessio Galatolo, Meriem Beloucif, Katie Winkle

Pretrained Language Models (PLMs) store extensive knowledge within their weights, enabling them to recall vast amount of information. However, relying on this parametric knowledge brings some limitations such as outdated information or gaps in the training data. This work addresses these problems by distinguish between two separate solutions: knowledge editing and knowledge augmentation. We introduce Difference Injection for Efficient Knowledge Augmentation and Editing (DIEKÆ), a new method that decouples knowledge processing from the PLM (LLaMA2-7B, in particular) by adopting a series of encoders. These encoders handle external knowledge and inject it into the PLM layers, significantly reducing computational costs and improving performance of the PLM. We propose a novel training technique for these encoders that does not require back-propagation through the PLM, thus greatly reducing the memory and time required to train them. Our findings demonstrate how our method is faster and more efficient compared to multiple baselines in knowledge augmentation and editing during both training and inference. We have released our code and data at https://github.com/alessioGalatolo/DIEKAE.

Alessio Galatolo, Zhenbang Dai, Katie Winkle et al.

Fine-tuning Large Language Models (LLMs) with first-order methods like back-propagation is computationally intensive. Zeroth-Order (ZO) optimisation uses function evaluations instead of gradients, reducing memory usage, but suffers from slow convergence in high-dimensional models. As a result, ZO research in LLMs has mostly focused on classification, overlooking more complex generative tasks. In this paper, we introduce ZOPrO, a novel ZO algorithm designed for Preference Optimisation in LLMs. We begin by analysing the interplay between policy and reward models during traditional (first-order) Preference Optimisation, uncovering patterns in their relative updates. Guided by these insights, we adapt Simultaneous Perturbation Stochastic Approximation (SPSA) with a targeted sampling strategy to accelerate convergence. Through experiments on summarisation, machine translation, and conversational assistants, we demonstrate that our method consistently enhances reward signals while achieving convergence times comparable to first-order methods. While it falls short of some state-of-the-art methods, our work is the first to apply Zeroth-Order methods to Preference Optimisation in LLMs, going beyond classification tasks and paving the way for a largely unexplored research direction. Code and visualisations are available at https://github.com/alessioGalatolo/VisZOPrO

Katie Winkle, Emmanuel Senft, Séverin Lemaignan

Participatory Design (PD) in Human-Robot Interaction (HRI) typically remains limited to the early phases of development, with subsequent robot behaviours then being hardcoded by engineers or utilised in Wizard-of-Oz (WoZ) systems that rarely achieve autonomy. We present LEADOR (Led-by-Experts Automation and Design Of Robots) an end-to-end PD methodology for domain expert co-design, automation and evaluation of social robots. LEADOR starts with typical PD to co-design the interaction specifications and state and action space of the robot. It then replaces traditional offline programming or WoZ by an in-situ, online teaching phase where the domain expert can live-program or teach the robot how to behave while being embedded in the interaction context. We believe that this live teaching can be best achieved by adding a learning component to a WoZ setup, to capture experts' implicit knowledge, as they intuitively respond to the dynamics of the situation. The robot progressively learns an appropriate, expert-approved policy, ultimately leading to full autonomy, even in sensitive and/or ill-defined environments. However, LEADOR is agnostic to the exact technical approach used to facilitate this learning process. The extensive inclusion of the domain expert(s) in robot design represents established responsible innovation practice, lending credibility to the system both during the teaching phase and when operating autonomously. The combination of this expert inclusion with the focus on in-situ development also means LEADOR supports a mutual shaping approach to social robotics. We draw on two previously published, foundational works from which this (generalisable) methodology has been derived in order to demonstrate the feasibility and worth of this approach, provide concrete examples in its application and identify limitations and opportunities when applying this framework in new environments.

Alan F. T. Winfield, Katie Winkle

Risk Assessment is a well known and powerful method for discovering and mitigating risks, and hence improving safety. Ethical Risk Assessment uses the same approach but extends the envelope of risk to cover ethical risks in addition to safety risks. In this paper we outline Ethical Risk Assessment (ERA) and set ERA within the broader framework of Responsible Robotics. We then illustrate ERA with a case study of a hypothetical smart robot toy teddy bear: RoboTed. The case study shows the value of ERA and how consideration of ethical risks can prompt design changes, resulting in a more ethical and sustainable robot.

Alan F. T. Winfield, Katie Winkle, Helena Webb et al.

Robot accidents are inevitable. Although rare, they have been happening since assembly-line robots were first introduced in the 1960s. But a new generation of social robots are now becoming commonplace. Often with sophisticated embedded artificial intelligence (AI) social robots might be deployed as care robots to assist elderly or disabled people to live independently. Smart robot toys offer a compelling interactive play experience for children and increasingly capable autonomous vehicles (AVs) the promise of hands-free personal transport and fully autonomous taxis. Unlike industrial robots which are deployed in safety cages, social robots are designed to operate in human environments and interact closely with humans; the likelihood of robot accidents is therefore much greater for social robots than industrial robots. This paper sets out a draft framework for social robot accident investigation; a framework which proposes both the technology and processes that would allow social robot accidents to be investigated with no less rigour than we expect of air or rail accident investigations. The paper also places accident investigation within the practice of responsible robotics, and makes the case that social robotics without accident investigation would be no less irresponsible than aviation without air accident investigation.