John Grundy

A critical challenge in Computer Science Higher Education (CSHE) is gender disparity. Despite ongoing efforts, gender disparity remains a crucial challenge in CSHE, affecting student engagement and learning outcomes. We wanted to investigate CSHE educators’ views on gender inclusivity or lack thereof, key factors influencing inclusivity in CSHE settings, and how various student facets, characteristics and behaviours, as perceived by education, influence different gender student participation and engagement. We used semi-structured interviews (n=20) with a range of experienced CSHE educators to explore these factors in depth. We report on our key findings, recommendations and future research directions.

The present study used event-related potentials to examine how bilingual experiences shape neural responses to contextual cues and conflict resolution. Fifty bilingual undergraduates completed a context-specific proportion congruency (CSPC) Stroop task while electroencephalography was recorded. Greater L2 usage was associated with reduced conflict effects, consistent with more efficient control. The N2 was larger for incongruent than congruent trials and varied by context, with lower L2 proficiency associated with greater context-dependent increases in N2 amplitude, whereas higher proficiency was associated with more stable responses. The conflict slow potential showed that earlier L2 acquisition was associated with greater sustained attentional resource availability across contexts. Together, these findings indicate that processing efficiency is reflected in how control demands are distributed across early and late stages of processing. More broadly, bilingual experiences shape distinct neural control profiles, highlighting the importance of interpreting behavioral and neural measures jointly.

Statically typed languages offer many advantages in software engineering, including bug prevention, enhanced code quality, and reduced maintenance costs. However, these benefits come at the expense of a steep learning curve and a slower development pace. Although known for its expressive and strong type system, Haskell often frustrates programmers as they learn and use that type system, especially when debugging type errors. We introduce Goanna, a novel type checker for Haskell that focuses on improving error diagnostics. When a type error occurs, Goanna identifies a comprehensive list of possible causes and provides a potential course of action for each. To achieve this, Goanna uses constraint logic programming and Minimal Correction Subsets (MCSes) to support its diagnostics. We evaluated Goanna’s effectiveness on 49,750 Haskell programs, demonstrating its ability to identify root causes with higher accuracy (81.6%) than traditional tools (Helium, 65.5%; GHC, 33.7%). In our benchmark study, Goanna shows performance constraints when diagnosing large programs containing complex errors, but remains responsive enough to provide real-time debugging assistance for small to medium-sized programs.

Context: Major healthcare-AI frameworks set out high-level ethical principles but say little about how to implement them. Objectives: We ask whether everyday online discussion can reveal the practical, deployment-level concerns that formal elicitation tends to miss, and whether those concerns can suggest requirements. Methods: We collected 10,212 Reddit threads from 33 healthcare and technology communities (January 2022 to August 2025) and annotated each with a multi-agent LLM pipeline for self-presented stakeholder participation, ethical relevance, and five ethical dimensions. Majority voting retained 2,597 ethics-relevant thread-stakeholder instances. Human review of a stratified 200-instance sample agreed with 77.5% of the positive stakeholder attributions and 97.5% of the ethical-relevance decisions, but only 66.0% of the dimension labels; we therefore read the dimension frequencies as rough indications rather than population estimates. Results: In this corpus, Safety and Accountability came up most often and usually together. The manually reviewed threads surfaced concrete issues such as correlated liability when one model is shared across sites, the lack of a human-performance baseline for judging safety, and bias introduced by site-specific data. Conclusion: We treat Reddit as a source of naturally occurring, self-presented discussion, not as a representative sample of any profession. Its value is in surfacing issues early: the concerns it reveals give requirements teams concrete candidates to test and confirm through direct engagement with stakeholders.

Context: Cultural inclusiveness remains under-examined in software usability evaluation, despite culture’s strong influence on how users interpret information and engage with digital systems.Objective: This paper presents CultureMag, a preliminary approach that extends the InclusiveMag method to identify culture-related usability issues in software interfaces.Method: We applied CultureMag in the context of parent-teacher digital communication in Australia and on two widely used school communication platforms, Compass and Seesaw. To operationalise cultural diversity, we developed two complementary persona-construction methods: an empirical method grounded in survey data collected from parents and a theory-based method derived from Hofstede’s cultural dimensions, contextualised for parent-teacher communication. We then applied both persona sets in structured cognitive walkthroughs of the platforms to identify cultural inclusiveness issues.Results: The findings show that CultureMag can reveal usability issues of an interface. We also found that survey-based personas were more directly useful in the walkthroughs, while theory-based personas provided broader conceptual coverage but required additional interpretation.Conclusion: CultureMag offers a preliminary yet repeatable approach to incorporating culture into usability evaluation. The findings suggest that culturally informed personas can help developers identify issues that conventional usability methods may miss, however, further studies are needed across other software contexts.

Virtual Reality (VR) offers immersive, experiential learning experiences in computing education, however, its successful integration in classrooms require navigating a range of practical organizational and pedagogical challenges. This poster presents insights drawn from post-hoc educator interviews, and researcher reflections, who collaboratively deployed an 'in-the-wild' VR-based lesson in a higher education IT Professional Practice course. Our findings highlight several planning, logistical and class orchestration aspects that influence VR integration, rarely captured in controlled VR studies. We also draw attention to the importance of research aspects such as establishing partnerships with teaching teams, being flexible with classroom dynamics, educator safety and iterative refinements for sustainable VR integration. These findings can provide practical guidance for computing educators and researchers seeking to implement or study VR in authentic contexts.

How software developers interact with Artificial Intelligence (AI)-powered tools, including Large Language Models (LLMs), plays a vital role in how these AI-powered tools impact them. While overreliance on AI may lead to long-term negative consequences (e.g., atrophy of critical thinking skills); underreliance might deprive software developers of potential gains in productivity and quality. Based on twenty-two interviews with software developers on using LLMs for software development, we propose a preliminary reliance-control framework where the level of control can be used as a way to identify AI overreliance and underreliance. We also use it to recommend future research to further explore the different control levels supported by the current and emergent LLM-driven tools. Our paper contributes to the emerging discourse on AI overreliance and provides an understanding of the appropriate degree of reliance as essential to developers making the most of these powerful technologies. Our findings can help practitioners, educators, and policymakers promote responsible and effective use of AI tools.