Human-Computer Interaction practitioners have been proposing best practices in user interface design for decades. However, generative Artificial Intelligence (GenAI) brings additional design considerations and currently lacks sufficient user guidance regarding affordances, inputs, and outputs. In this context, we developed a recommender system to promote responsible AI (RAI) practices while people prompt GenAI systems. We detail 10 interviews with IT professionals, the resulting recommender system developed, and 20 user sessions with IT professionals interacting with our prompt recommendations. Results indicate that responsible prompting recommendations have the potential to support novice prompt engineers and raise awareness about RAI in prompting-time. They also suggest that recommendations should simultaneously maximize both a prompt’s similarity to a user’s input as well as a diversity of associated social values provided. These findings contribute to RAI by offering practical ways to provide user guidance and enrich human-GenAI interaction via prompt recommendations.

Human-Computer Interaction practitioners have been proposing best practices in user interface design for decades. However, generative Artificial Intelligence (GenAI) brings additional design considerations and currently lacks sufficient user guidance regarding affordances, inputs, and outputs. In this context, we developed a recommender system to promote responsible AI (RAI) practices while people prompt GenAI systems, by recommending addition of sentences based on social values and removal of harmful sentences. We detail a lightweight recommender system designed to be used in prompting-time and compare its recommendations to the ones provided by three base large language models (LLMs) and two LLMs fine-tuned for the task, i.e., recommending inclusion of sentences based on social values and removal of harmful sentences from a given prompt. Results indicate that our approach has the best F1-score balance in terms of recommendations for additions and removal of sentences to promote responsible prompts, while a fine-tuned model obtained the best F1-score for additions, and our approach obtained the best F1-score for removals of harmful sentences. In addition, fine-tuned models improved the objectiveness of responses by reducing the verbosity of generated content in 93% when compared to the content generated by base models. Presented findings contribute to RAI by showing the limits and bias of existing LLMs in terms of recommendations on how to create more responsible prompts and how open-source technologies can fill this gap in prompting-time.

Societal implications of technology are often considered after public deployment. However, broader impacts ought to be considered during the onset and throughout development to reduce potential for harmful uses, biases, and exclusions. There is a need for tools and frameworks that help technologists become more aware of broader contexts of their work and engage in more responsible and inclusive practices. In this paper, we introduce an online card tool containing questions to scaffold critical reflection about projects’ impacts on society, business, and research. We present the iterative design of the Responsible & Inclusive Cards and findings from five workshops (n=21 participants) with teams distributed across a multinational technology corporation, as well as interviews with people with disabilities to assess gameplay and mental models. We found the tool promoted discussions about challenging topics, reduced power gaps through democratized turn-taking, and enabled participants to identify concrete areas to improve their practice.

Abstract Computer anxiety (CA) can be defined as fear and worries that someone may feel when using computers. Thus, people with CA may face problems when using computers at home, at work or for study purposes, resulting in multiple forms of barriers even before the actual interaction with computers. In this context, the purpose of this research is to identify user interface elements impacting task performance (i.e. distractors) for people with CA, using mouse fixation analysis as a proxy for eye gaze data. The study explores the relationship of mouse and eye gaze data collected with the help of 39 older adults interacting with a website. Results show that it is possible to identify UI elements acting as distractors (e.g. carousel, top menu) as well as those with which people with CA faced problems (e.g. side menu, search box, map), based on mouse fixations. Moreover, statistical differences show that the number of mouse fixations in navigation, content and distractors is different for different levels of CA. Furthermore, differences were found between CA groups regarding mouse and eye fixations, indicating that participants with higher CA levels had difficulty differentiating which areas of interest they should interact with using mouse. From the results, one expects that personalized systems could use the proposed approach to identify UI elements acting as distractors using mouse data and then simplify UIs based on different levels of CA.

Designing systems to offer support to experts during cognitive intensive tasks at the right time is still a challenging endeavor, despite years of research progress in the area. This paper proposes a gaze model based on eye tracking empirical data to identify when a system should proactively interact with the expert during visual inspection tasks. The gaze model derives from the analyses of a user study where 11 seismic interpreters were asked to perform the visual inspection task of seismic images from known and unknown basins. The eye tracking fixation patterns were triangulated with pupil dilations and thinking-aloud data. Results show that cumulative saccadic distances allow identifying when additional information could be offered to support seismic interpreters, changing the visual search behavior from exploratory to goal-directed.