Human oversight of artificial intelligence systems is increasingly mandated by regulation, yet empirical evidence on which interface design choices actually improve oversight quality remains scarce. This thesis investigates how two modifiable design factors, information signal granularity and intervention option range, affect the effectiveness and perceived workload of human overseers in an AI-assisted fraud detection task. A 3×3 between-subjects experiment was conducted online via Prolific (N = 144), in which participants reviewed 30 bank account applications flagged by a Gradient Boosting model under one of nine conditions, crossing three levels of information signal (plain case features, categorical risk level indicator, and continuous model confidence score) with three levels of intervention options (binary decision, decision with delegation, and decision with flagged delegation). Oversight effectiveness was operationalised as a composite score rewarding correct classifications and appropriate delegation decisions, and penalising both misclassifications and over-delegation. Perceived workload was measured using the NASA Task Load Index. Self-reported domain expertise was included as a covariate. Neither main effect reached the pre-registered Bonferroni-corrected significance threshold of α = 0.008, and no significant interaction was found on either dependent variable. A marginal effect of information signal granularity on oversight effectiveness was observed (F(2, 134) = 3.29, p = .040, η²p = .047), with a non-monotonic pattern in which the categorical risk level indicator outperformed both the plain information baseline and the continuous model confidence score. This reversal of the hypothesised ordering suggests that, for non-expert overseers, a well-designed categorical signal may be more actionable than a continuous probability score, as interpreting the latter requires complementary domain knowledge not uniformly present in a general population. The results are treated as preliminary, since the study was underpowered relative to the pre-registered target of N = 288, due to the planned expert-screened wave not being completed within the thesis timeline. Theoretical and practical implications for the design of human oversight interfaces are discussed, with particular attention to the relationship between signal granularity and overseer expertise. ... Read more

This graduation project investigates how conversational AI can reshape interactions within smart homes by making energy trade-offs visible, felt, and discussable. While most smart home systems prioritise seamless automation and passive convenience, this work challenges that model by introducing devices that resist, reflect, and negotiate. Drawing on speculative design, animistic interaction, and sustainable HCI, the project reimagines smart devices not as silent servants but as expressive co-inhabitants.

Using a Research through Design (RtD) approach, the project developed four prototypes that tested different configurations of personality, memory, design friction, and conversational flow. These prototypes explored how smart devices could express opinions, question user requests, and make the cost of comfort more tangible. The final concept, Talkback Home, features three characterful bots—HeaterBot, LampBot, and KettleBot—who respond to user requests via Telegram with emotional tone, value-based reasoning, and concrete trade-offs linked to energy constraints.

The system integrates conversational AI with tangible feedback (via NeoPixels and servo motors) to link dialogue with real-world consequences. A Figma-based test with six participants revealed that people found the system more engaging when devices had strong personalities, humour, and referenced everyday habits. However, users also valued agency and clear override options.

The project contributes to design research by showing how expressive AI systems can encourage reflection on energy use, make value conflicts visible, and promote ethical engagement in domestic settings. It proposes an alternative vision of the smart homeone that invites negotiation, care, and accountability instead of passive automation. ... Read more

Conversational agents (CAs) are increasingly being adopted across various domains, with enterprises particularly interested in leveraging these technologies to explore new opportunities. CAs, such as Copilot, are transforming human-computer collaboration in the workplace by serving as effective work assistants. Additionally, their advanced natural language understanding capabilities position them as powerful tools for facilitating crowdsourcing within enterprises.
This thesis investigates the dual role of CAs in enterprises: as work assistants for employees and as facilitators of crowdsourcing. By acting as work assistants, CAs interact with a broad range of employees, making them well-suited to gather valuable insights that contribute to the company’s knowledge base and support collaborative problem-solving. The central research question explores how to effectively engage and motivate employees in this crowdsourcing process through CA personality design, which is relatively unexplored. Drawing on the "Computers Are Social Actors" (CASA) framework by Nass et al. (1994), this thesis examines CA personality design, aiming to enhance human-CA interaction by eliciting social responses from employees and thereby increasing their engagement.
We approached these research questions from three perspectives: (1) designing and specifying CA personalities for effective implementation in enterprises, (2) identifying the most and least promising qualities of CA personality designs in this context, and (3) designing an empirical study within the enterprise to measure which personality archetype design is the most effective.
We reviewed literature on the context of conversational crowdsourcing in enterprise settings, and CA design methods with a focus on CA anthropomorphism. These findings informed the initial design phase, where CA personality archetypes were tested in an enterprise environment. Based on these insights, we developed design guidelines for enterprise CA personality design. After refining the archetypes, we proposed three unique CA personalities aimed at enhancing employee participation in enterprise crowdsourcing through their interactions with CAs. These archetypes were then implemented in the enterprise's internal chat platform for an experiment, where employee interactions with the CAs were collected and analysed both quantitatively and qualitatively.
The results of our experiment reveal how different CA personalities impact user engagement in enterprise crowdsourcing, demonstrating how distinct CA personalities can be effectively translated into LLM prompts and perceived by users. The study also highlights the relationship between CA personality and the triggering of users' social responses, as observed through sentiment analysis. Additionally, our research identifies the challenge of balancing social engagement with maintaining professionalism in enterprise CAs to build user trust.
This thesis concludes with a discussion of the findings and their implications for future research in enterprise CA design.
... Read more

Several useful technologies to help people with dementia already exist. Robotic pets, for example, have been shown to alleviate negative symptoms associated with dementia. Most robotic pets, however, only have pet-like functionality. This is useful on its own, but adding additional care functionality would be helpful. A different useful tool for people with dementia is reminder systems. These systems can help them remember to do important daily tasks, like eating meals and taking medicine. These systems, however, usually use a phone as a medium. The user might leave their phone in another room, and if they do not take it to the location where the task is to be completed, they cannot check what the reminder was for.

This thesis aims to combine these two systems, and presents the design of an integrated robot reminder system for people with dementia who live at home. It aims to address the problem of people with dementia forgetting important things, improve their quality of life, and reduce the stress experienced by their informal caregivers. The system consists of an animal-like robotic pet, which delivers reminders for important activities of daily living directly to the user with dementia. It guides them to a location in the house, where a screen tells them what the reminder is for. Through sensors placed in the house, the system can automatically detect whether some reminders have been completed, and can use contextual clues to send reminders at the best time. The research question for this thesis is “How can an integrated robotic reminder system effectively support people with dementia living at home in completing daily tasks while alleviating stress of their informal caregiver?”

Three research topics are explored in the thesis. The first topic is that of value-sensitive design. The two main user groups, people with dementia and caregivers, both have values they want to see reflected in the system. Sometimes, these values may clash. The conflict between these values is explored in this thesis. Secondly, the interaction between the robot and the person in the context of a reminder system is researched. It should be clear to a user that the robot is trying to convey a reminder. The third topic relates to the software architecture and software engineering requirements. Private data should be stored securely, and the requirements should be written to make future development easy. Recommendations are made regarding the ideal setup of the system, ensuring proper security and usefulness. These technical choices were validated and improved through two interviews with experts in the field of privacy.

The system was designed through an iterative process. In order to improve the system and validate its acceptability, two design workshops were held, one with professional caregivers and one with people with dementia living in a care home. We showed them a high-fidelity robot prototype, and showed the caregivers a user interface prototype for a caregiver app. These workshops showed that the combination of the robot and the reminder functionality was appealing, though not everyone was as interested. ... Read more

Space exploration has historically driven technological innovation, resulting in significant advancements with applications in various industries on Earth. The exploration of Mars has emerged as a crucial objective, offering opportunities to search for signs of life and gain insights into planetary evolution. However, working on Mars presents numerous challenges, including the lack of a breathable atmosphere, different gravity, extreme temperatures, and the need for infrastructure development. Robots have proven resilient on the Martian surface but still require real-time control and decision-making from ground operators. To address this challenge, a promising solution involves astronauts in orbit around Mars controlling the robots, utilizing high-bandwidth communication techniques and autonomous capabilities.
This work focuses on the Surface Avatar project, led by the DLR and ESA, which involves the humanoid robot, Rollin Justin. The project aims to gain valuable insights into the efficient control of robots in future space missions, particularly through collaborative exploration and construction tasks. Rollin Justin, equipped with autonomous capabilities, features a user interface that allows manual controls through various input devices and autonomous operation through interface commands.
While the concept of an astronaut-robot pairing shows promise, several challenges remain. Error handling during teleoperation poses a significant issue, as error messages often lack specificity, leaving astronauts confused and without immediate assistance due to the distance between Earth to Mars and the associated communication delays. Limited situational awareness, unfamiliarity with robot constraints, and a large time gap between training and usage further complicate astronaut interactions with the robot. Addressing these problems is critical for optimizing astronaut-robot cooperation and reducing cognitive workload during Mars missions.
To address the challenges, this work adopts a research-through-design approach, specifically focusing on user experience research and design. Extensive initial research including sessions at the DLR and literature review, was conducted to identify key issues impacting error-handling capabilities. Based on the research findings, conceptual solutions were developed to address the identified core issues. These concepts were evaluated for feasibility and desirability, considering expert input. Selected concepts were further developed, drawing inspiration from game cues and elements for user interface design. High-fidelity prototypes were created to represent the refined concepts accurately: A third-person perspective including game elements to allow for better situational awareness and a debug page that guides the user through potential error reasons in the moment of an occurring planning error. The prototypes underwent evaluation using various methods, including user sessions at the DLR and a comparative study. 
The results for both prototypes reveal important enhancements in user experience and a reduction in cognitive workload compared to the existing system. The findings led to informed recommendations for further improvements in the interface design, the robot’s camera setup and the communication of errors to enhance error-handling capabilities for astronauts in future missions.  ... Read more

Sound in UX design can be used in a variety of ways; to navigate the interaction, for branding, to set the ambience etc. The importance of sound for the user experience of a product is undeniable; however, this is not represented in the practices of UX designers. UX designers lack the tools and knowledge needed to integrate sound design into their prototyping activities. As a result, the sound is often considered last-moment, limiting the creative potential and added value of sound in the interaction. This report explores the possibilities of integrating sound design activities into the practices of UX designers. This is done through design research activities such as interviews, co-creation sessions, creating (interactive) prototypes and user testing. The aim of the project is to provide UX designers with prototyping tools for sound design. Common sound design tools are difficult to integrate in prototyping activities as UX designers often lack the skills and knowledge needed to operate these tools in a quick-paced and iterative way; which is needed in order to facilitate prototyping. Furthermore ways of bridging the semantic gap between the stakeholders is explored as well as the potential of tangible interfaces in making sound design more intuitive and engaging for UX designers. Finally, a concept is introduced: The Timbreworld, parts of which are embodied in an interactive prototype. The Timbreworld is a (tangible) interface for sonic sketching, it enables its users to create quick and iterative sound prototypes. The interactive prototype has been tested with participants and evaluated based on the design guidelines provided in the report ... Read more

The Dutch Government has set an ambitious goal for Amsterdam Zuidoost to become energy neutral by 2040. While technological advancements and infrastructure developments are progressing, the current transition risks widening social inequalities and leaving vulnerable groups behind. Thus, an inclusive transition, which recognises that not everyone can afford to invest in an energy-sustainable future, is crucial. In a bid to accelerate the transition, the Local Inclusive Future Energy (LIFE) Project is, amongst others, working on utilising local and sustainable energy to ensure the inclusion of local residents in the low-income district of Amsterdam Zuidoost.

In the context of the energy transition, inclusion typically refers to participation in energy exchange. Energy exchange is taking shape through the development of local, decentralised energy systems. Emerging in neighbourhoods and based on Peer-to-Peer and Transactive Energy models, these Local Energy Market Systems (LEMS) enable local production, storage and distribution of energy. However, designed to prioritise the interests of the grid, LEMS frame energy exchange solely as a transaction or trade, subjecting residents to the roles of profit-motivated buyers and sellers and thereby overshadowing the potential for fostering active inclusion and creating social value. With its unique mix of entertainment venues, social housing and large renewable asset owners, Amsterdam Zuidoost poses unique challenges and opportunities for becoming a pioneer in a socially inclusive transition.

To move beyond the rational and market-driven constraints which govern current energy exchanges, this project takes a new anthropological approach, framing it as a primarily social, community-based and relational practice.

By means of extensive ethnographic and research-through-design activities, the potential for socially inclusive energy exchange in Amsterdam Zuidoost is explored. Characterised by tight-knit social silos, residents stick within their own circles and thus social cohesion and a sense of community is lacking. Whilst entrepreneurial behaviour is evident, it remains behind doors and the skills and capabilities of local neighbourhoods is hidden and untapped.

To tap into the overlooked potential of local expertise and to expand social networks, this project suggests a new type of exchange, coined as ‘relational giving’. Through the giving of services and actions in return for local tokens opportunity is created for the formation and strengthening of social relationships. ‘Relational giving’ translates in the concept of ‘energy actions’: local energy related actions such as handy work and education. Local tokens are introduced as a tool to support relational giving, incentivising energy actions, recognising contributions and, importantly, facilitating exchange between the socially distant.

The vision for relational energy exchange is told through the mix media outputs of this project. Besides video, an energy actions enactment game is designed to activate stakeholders and a framework for a ‘Social Local Relational Energy Ecosystem’, which considers the roles of multiple actors in unlocking social energy, is presented.

Ultimately, this project aims to inspire and offers a new perspective on energy exchange, introducing tangibility to an entangled and unexplored space. Energy is repositioned not as a commodity to be bought and sold, but as an agent for building inclusive and socially cohesive communities.
... Read more

As robots become increasingly popular and co-inhabit spaces with people, there is a need for designing robots that people are comfortable interacting with. For robots to successfully integrate into human society, it is important to design robots in a way that enhances the quality of the human-robot interaction. While the human-robot relationship typically exists to serve people, there will be instances when robots require human assistance. This project aims to investigate people’s interactions with robots that need their assistance. It employs a robot designed to test people’s willingness to help it achieve a goal under various scenarios.
The project in its entirety was quite an ambitious one given the limited scope of a graduation project. A literature review carried at the beginning helped identify key features a robot and an interaction should have, in order for it to be perceived positively by humans interacting with it. This review culminated in a list of design considerations which informed the design and the interaction. These design considerations were referred to throughout the project, while choosing ideas for the interaction and while detailing the chosen scenario. A state machine was used to explain the general behavioural states of the robot, and the electronic components that facilitated the desired interaction were chosen. These components were coded to reflect the desired behaviour of the robot during the interaction. Alongside this process, the robot was embodied as sub-assemblies incorporating these components. Once the robot was given a form, the components were designed to fit within a modular body that accounted for zoomorphic and anthropomorphic features, while retaining functionality of the components inside. During its embodiment, the robot was given the ability to express itself through displayed speech, sounds, facial expressions and body language. Additional structures like "nests" were also designed to support the robot's interaction. Once the interaction was designed and embodied, the robot was then plot tested, before being used as a research probe to conduct human-robot interaction (HRI) experiments to address the research questions outlined at the start of the project. Data was gathered through data logging via the robot, user observations and interviews. The results of the experiments indicate that people's willingness to help a robot is influenced by the narrative of a human-robot interaction, the effort level of the task required to help the robot, as well as individuals' attitudes towards robots. Increasing the effort level reduced people's willingness to help the robot, but only when the narrative of the interaction was simple and functional, demonstrating that a narrative which tries to engage emotionally with participants has a positive effect on their altruistic response. The experiments were also able to shed some light on the social dynamics that emerge when a robot co-inhabits a space with people, as more people informed others about the robot.
... Read more

The development of intelligent systems has created enormous opportunities to improve and change human lives. These systems are heavily reliant and driven by data and algorithms to achieve optimal user experiences for myriad users. However, these data-driven products/services are multi-dimensional and multifaceted and do not necessarily have the same meaning and value to all its users. Different users may have different intentions of use for the system. Each user could also define different goals that they may want to achieve using the system. Additionally, what goals the organization has for the users might not align with what the end users what for themselves. These differences in intentions are called as multi-intentionality. In such scenarios, traditional design enforces the idea of simplifying interfaces that frame or dictate certain intentions of use for people. But this can be considered a sensitive issue because the end-users are unaware of these other potentially conflicting intentions. This creates an increase in tension between intelligent systems and the needs of end-users but also a sense of mistrust. Hence, there is a need to create a sense of legibility to the users on the behavior of these systems and the other intentions of use in order to enable trust. In this project, a way to help users sustainably consume energy is explored through the lens of multi-intentionality. The main aim of the research is to explore the meaning of intentions and multi-intentionality in the given context. Following this, the goal becomes to concretely represent the multi-intentionality into something more tangible for the users and applicable to the real-world context. The initial research resulted in identifying intention profiles for users, the intentionality gap, and the need to capture intentions that is required in order to bring more legibility and transparency to a system that the user might interact with. Through multiple iterations, a digital interface is created that through various communication data, data visualizations and recommendation designs tries to bridge the gap of intentionality between the two stakeholders (user and energy provider). By capturing intentions, portraying the tradeoffs of user choices, showcasing their energy use, presenting their energy profile with respect to their goals, designing transparent recommendations an effective proposal to bridge the gap between intentions is made. The designs were made with concrete thinking on how it could be applied to a real world problem and were validated through tests with users. ... Read more