This thesis explores the use of in-ear magnetosensing (EarMag) as a novel sensing technique for detecting jaw and head movements in the context of human-computer interaction. While prior work has explored the use of acoustic, inertial, and visual sensing, the potential of EarMag remains to be explored. As a proof of concept, 17 orofacial physiotherapy-related exercises were collected from 21 participants using EarMag-enabled earables. A soft voting ensemble of support vector machine and random forest achieved 76% accuracy for five exercises on an unseen test set of ten users. While individual anatomical differences pose challenges for generalization, this work highlights the potential of EarMag for applications such as assistive technologies, silent speech interfaces, and biosignal tracking.

Recent advances in LLMs have transformed AI coding assistants from simple autocompletion tools into conversational partners that support a wide range of development tasks through natural language interaction. While this shift promotes closer human-AI collaboration, this also places a burden on developers to craft precise prompts and integrate sufficient context to accomplish their task.
To address this, there is growing interest in proactive AI systems that can anticipate developer intent and surface timely, contextually relevant suggestions - potentially revealing insights developers might not have considered.

However, existing research on proactive coding assistance has largely been confined to controlled, experimental settings, leaving open questions about its effectiveness and integration in real-world development environments.

In this work, we design and implement a proactive AI assistant within JetBrains Fleet, a polyglot enterprise IDE. Our system signals AI activity via in-editor cues and delivers context-aware code improvement suggestions via a chat interface. We adopt heuristics based on IDE activity to determine when it is timely to intervene, during various stages across the development workflow: formulating needs, idea execution and committing changes.
Lastly, we conducted a five-day in-the-wild study with 18 professional developers to evaluate its impact on user experience and engagement patterns.

Our findings demonstrate that timing is a critical aspect influencing perceived value of proactive AI assistance.
Suggestions delivered at natural workflow boundaries (e.g. post-commit) achieved the highest engagement, while mid-task interventions were often dismissed.
The consistent engagement trends (and voluntary continued use) indicate strong potential for integrating proactive AI into everyday software engineering workflows.
Our study not only underscores the feasibility and practical value of proactive assistance in enterprise development environments, but also provides actionable design insights for future tools.
While challenges remain - such as improving suggestion relevance and supporting diverse user preferences - these highlight important directions for advancing adaptive, intent-aware proactive systems that can further enhance developer workflows.

Large Language Models (LLMs) have shown impressive performance in various domains, including software engineering. Code generation, a crucial aspect of software development, has seen significant improvements with the integration of AI tools. While existing LLMs have show very good performance in generating code for everyday tasks, their application in industrial settings and domain-specific contexts remains largely unexplored. This thesis investigates the potential of LLMs to generate code in proprietary, domain-specific environments, with a specific focus on the leveling department at ASML. The primary goal of this research is to assess the ability of LLMs to adapt to a domain they have not encountered before and to generate complex, interdependent code in a domain-specific repository. This involves evaluating the performance of LLMs in generating code that meets the specific requirements of ASML. To achieve this, the thesis investigates various prompting techniques, compares the performance of generic and code-specific LLMs, and examines the impact of model size on code generation capabilities. To evaluate the code generation capabilities of LLMs in repository-level scenarios, we introduce a new performance metric, build@k, designed to measure the effectiveness of generated code in compiling and building projects. The results showed that both prompting techniques and model size have a substantial influence on the code generation capabilities of LLMs. However, the performance difference between code-specific and generic LLMs was less pronounced and varied substantially across different model families.

Delayed software projects are one of the biggest threats to the integrity of many project portfolios. If portfolio managers were able to foresee delays, they could better manage risks, make adjustments to the planning and reduce delay propagation. In their 2023 paper "Dynamic Prediction of Delays in Software Projects using Delay Patterns and Bayesian Modeling", Kula et al. propose an AI solution for the problem of ineffective delay prediction of software projects. Even though Kula et al. achieved positive results, they are bound to ING’s data, and thus may not be representative of software projects in other companies or industries. This thesis builds on Kula et al.’s work by applying the same methodology to a new dataset - Coca-Cola Hellenic's Project Portfolio. By doing so, it assesses the robustness and generalisability of Kula et al.'s delay prediction model. The results clearly indicate that the model was unsuccessful at Coca-Cola Hellenic, as it proved no better than random guessing.
Differences in dataset size and quality were identified as the primary cause for the lack of performance. Furthermore, contextual factors were likely a major contribution to the difference in results, namely differences in industry, organisational structure and agile maturity. These findings are valuable to anyone attempting to replicate this solution, or to organisations aiming to adopt AI-powered analytics.
Future research directions are suggested, such as a requirement framework for AI solutions and further replication of Kula et al.'s work in different contexts.

This research explores the integration of episodic memory and visual short-term memory in social robots to enhance engagement and interaction with older adults, particularly those with dementia, through art activities such as painting. Dementia significantly impacts millions globally, being one of the leading causes of death worldwide. Furthermore, it creates a substantial burden on caregivers and healthcare systems. This research aims to leverage social robots to provide meaningful engagement, companionship, and emotional support to individuals with dementia.

The study involved developing a painting application using Android Studio and a conversational agent using generative AI. The AI model was enhanced with episodic memory and visual short-term memory, enabling it to reference past interactions through episodic memory and analyze painting progress through recent visual inputs with short-term memory. This qualitative study included 8 participants, and post-session interviews were conducted to gain deeper insights into the participants' experiences with the system.

The study suggests that memory-enhanced social robots can improve engagement and interaction quality among older adults interested in painting, but have limited impact on those uninterested in the activity. Mixed responses were observed regarding the robot's role as a companion, with some participants feeling a sense of company while others did not. Overall, participants did not experience frustration due to the robot's presence.

Challenges include response time and real-time awareness, highlighting the need for a more intelligent system. Given the small sample size of eight participants, future research should involve larger groups for more comprehensive and reliable data.

This study explores the application of Large Language Models (LLMs) for eliciting tacit knowledge, a challenging area crucial for enhancing decision-making and innovation in organizations. Using a systematic literature review based on PRISMA workflow, the research assesses the potential of LLMs to bridge the gap between tacit knowledge and its articulation. Findings reveal that LLMs, with their advanced natural language processing capabilities, are suitable for capturing tacit knowledge that is typically inaccessible through traditional methods. This paper concludes that while LLMs hold potential for revolutionizing knowledge elicitation practices, careful consideration of their limitations and ethical use is essential. This research contributes to broader discussions on integrating AI in knowledge management and future directions to optimize LLMs’ utility in practical settings.

Trust is essential in human-AI collaboration, influencing efficiency, safety, and overall success. Communication influences trust, and the more effective it is, the greater the resulting trust and overall satisfaction. The research question addressed is: “How does a real-time visual (RTV) communication of the mental model of the agent’s trust affect the human teammate’s trust in the agent and overall satisfaction?”. Utilizing an independent measures design, 44 participants were divided into two groups: one experiencing explicit communication of the robot’s mental model, and a control group without such communication. Participants collaborated with an AI agent in a 2D grid-based Urban Search and Rescue game. The study combined subjective measures, such as surveys and open-ended questions, with objective measures. Findings indicate that real-time visual communication did not significantly enhance trust, but it did improve overall satisfaction.

To collaborate effectively, humans and AI agents need to trust each other. Communication between teammates is an essential component to achieve this, as it makes the AI system more understandable to humans. However, previous research lacks a focus on ways to build an AI agent's trust in its human teammate and, consequently, on how the AI's beliefs can be communicated to the human. As such, this study explores how real-time visual explanations of the AI agent's trust in its human teammate influence human trust and overall satisfaction. Through a user experiment (n=46) conducted on an Urban Search and Rescue simulation, integrating trust explanations was compared against a baseline containing no such information. Results show a statistically significant increase in both human trust and satisfaction when the explanations are provided, highlighting the need for further exploration into methods of communicating trust.

As technological capabilities progress, there is a growing imperative to enhance collaborative dynamics within human-agent teams. Artificial agents and humans possess capabilities that compensate for each other’s limitations. This paper outlines the effect of communication using real-time textual explanations of the artificial agents’ mental model of its trust in the human teammate. An experiment (n = 40) was conducted by examining the impact of real-time textual explanations of the artificial agents’ trust beliefs. The participants collaborated with an artificial agent during a search and rescue mission in a 2D grid world, where they had to rescue six victims in a 10 minute time frame. The results show that real-time textual explanations did affect the natural trust and satisfaction positively compared to the baseline method.

Human-agent teams (HATs) are becoming more prevalent in our current world, necessitating mutual trust between humans and machines. This trust is split into artificial trust (agents trusting humans) and natural trust (humans trusting agents). Both types must be facilitated for effective teamwork. It is hypothesized that communicating artificial trust effectively helps develop natural trust and overall satisfaction. A visual summary of explanations is proposed to serve as an effective communication method. Summaries allow for in-depth information processing, and visual representations are quicker to interpret. This paper examines the impact of using a visual summary of explanations to communicate the agent’s trust beliefs on the human teammate’s natural trust and overall satisfaction within HATs. An experiment (n=40) was conducted to study this effect. Participants collaborated with an artificial agent during an urban search and rescue operation in a simulated 2D grid-world environment. Results show that the inclusion of a visual summary increases the human teammate’s trust in the agent alongside their overall satisfaction. The paper emphasizes the need for further research with longitudinal studies to measure the long-term effectiveness of communicating artificial trust.

Human-agent teamwork (HAT) is becoming increasingly prevalent in fields such as search and rescue (SAR), where effective collaboration between humans and artificial agents is crucial. Previous studies have shown that trust plays a pivotal role in the success of HATs, influencing decision-making, communication, and potentially overall team performance.

This research investigates the impact of agent-provided explanations about the agent's trust in humans (artificial trust) and corresponding behavior changes on human trust in the agent and their satisfaction with explanations during a simulated SAR task. Two types of explanations were explored: Trust-Explained (TE) explanations, where the agent explains its trust level and trust-based decisions, and Trust-Unexplained (TU) explanations, which solely describe the agent’s behavior without reference to trust dynamics. Besides, this research also investigates the correlation between human trust and explanation satisfaction, and in the end, whether the differences in the provided explanations result in differences in team performance and artificial trust.

The study involved 40 participants divided into two groups: an experimental group (the trust-enhanced explanation group) receiving TE explanations and a control group (the non-trust explanation group) receiving TU explanations. Participants' trust in the agent, satisfaction with the explanations, and team performance and artificial trust were measured and analyzed. Contrary to initial expectations, no statistically significant differences in explanation satisfaction and human trust in the agent were found between the two groups. However, a strong positive correlation was observed between participants' satisfaction with the explanations and their trust in the agent, indicating that explanation quality plays a crucial role in human trust development. Furthermore, no significant differences in team performance were detected, suggesting that trust explanations may not directly influence task outcomes. In the analysis of artificial trust, the agent in the trust-enhanced explanation group exhibited more conservative adjustments in trust levels compared to the non-trust explanation group. This conservative approach may have influenced players in the trust-enhanced explanation group to adopt a more cautious or deliberate decision-making process, potentially prioritizing the comprehension of explanations over the optimization of task performance.

For future research, it may be worth delving deeper into the influence of trust explanations on user behavior, the more complex HAT task environments, the relationship between artificial trust and user behavior, the dynamic and adaptive explanations, and the causal relationship between explanation satisfaction and human trust in the agent to understand further how trust can be fostered in HAT.

As autonomous systems are increasingly integrated as a team member for collaborative tasks, trust in human-agent teams (HAT) becomes crucial to foster success. In many real world scenarios, trust violations are expected, thus demanding the use of trust repair strategies to restore damaged trust. Previous research has shown that expressing regret and providing explanations are effective strategies to rebuild human-agent trust.
However, the role of various team dynamics, such as interdependence relationships, remains unexplored. The aim of this paper is to examine the influence of interdependence levels on the effectiveness of trust repair strategies. To investigate this, an experiment was conducted in a collaboration environment with an urban search and rescue mission. Two interdependence conditions were introduced to analyse their effect on trust and collaboration fluency.
No significant evidence was found to support a relationship between interdependence and trust repair or collaboration fluency. However, as this study only considers two interdependence conditions, there is much more room for future work to explore further. This study can bring meaningful insights to design and facilitate agents that are more trustworthy in human-agent collaboration settings.

Interdependence relationships between humans and agents play a crucial role in the collaborative AI field. This research paper examines the impact of interdependence on trust violation, trust repair strategies, and collaboration fluency in human-AI teams. It compares independent cooperation and required interdependence approaches, focusing on collaborative AI, trust dynamics, and collaboration fluency. The paper presents a user study involving 30 participants in different interdependence conditions, analyzing data from trust surveys, collaboration fluency surveys, performance metrics, and AI agent idle time. The findings enhance understanding of human-agent collaboration fluency and as well as the interdependence relationship impact on trust.

Web3Recommend is a decentralized Social Recommender System implementation that enables Web3 Platforms on Android to generate recommendations that balance trust and relevance. Generating recommendations in decentralized networks is a non-trivial problem because these networks lack a global perspective due to the absence of a central authority. Further, decentralized networks are prone to Sybil Attacks in which a single malicious user can generate multiple fake or "Sybil" identities. Web3Recommend relies on a novel graph-based content recommendation design inspired by GraphJet, a recommendation system used in Twitter enhanced with MeritRank, a decentralized reputation scheme that provides Sybil-resistance to the system. By adding MeritRank's decay parameters to the vanilla Social Recommender Systems’ personalized SALSA graph algorithm, we can provide theoretical guarantees against Sybil Attacks in the generated recommendations. Similar to GraphJet, we focus on generating real-time recommendations by only acting on recent interactions in the social network, allowing us to cater temporally contextual recommendations while keeping a tight bound on the memory usage in resource-constrained devices, allowing for a seamless user experience. As a proof-of-concept, we integrate our system with MusicDAO, an open-source Web3 music-sharing platform, to generate personalized, real-time recommendations. Thus, we provide the first Sybil-resistant Social Recommender System, allowing real-time recommendations beyond classic user-based collaborative filtering. The system is also rigorously tested with extensive unit and integration tests. Further, our experiments demonstrate the trust-relevance balance of recommendations against multiple adversarial strategies in a test network generated using data from real music platforms.

Intelligent agents are increasingly required to engage in collaboration with humans in the context of human-agent teams (HATs) to achieve shared goals. Interdependence is a fundamental concept in teamwork. It enables humans and robots to leverage their capabilities and collaboratively work towards a shared goal, fostering the development of trust through joint activities. Considering the great importance of trust, the effectiveness of trust repair strategies is crucial as they help mitigate the negative consequences of errors, enabling efficient collaboration between humans and robots. For this reason, the effectiveness of the trust repair strategies must be examined comprehensively by taking into consideration multiple factors, including the interdependence relationships within HATs. This paper aims to examine the impact of a mix of interdependence and independence relationships on trust violation and repair, but also on collaboration fluency. Thus, an experiment (n = 30) was conducted to study how interdependence affects trust violation, trust repair and collaboration fluency. Participants collaborated with a robot during a search and rescue mission in a simulated environment. Results show that there is a significant influence of interdependence on trust violations, but not on collaboration fluency or trust repair. Furthermore, the paper also emphasises the need for future research that investigates the effectiveness of trust repair strategies for HATs in different interdependence relationships.

Nowadays, Human Autonomy Teams (HATs) are incorporated in many fields, where humans and autonomous agents work collaboratively to combine their capabilities with the ultimate goal of performing tasks more efficiently. In such environments, it is imperative to sustain a high level of trust between the agents as collaboration is not possible without mutual trust. Naturally, this implies that recovering trust following trust violation is also a crucial aspect of HATs. Moreover, besides team performance, the fluency of collaboration is another important factor to consider when evaluating the success of the teams. This paper aims to investigate the effect of opportunistic (soft) interdependence between the agents on trust violation, trust repair, and on collaboration fluency when compared against a baseline (complete independence) condition. In this paper, interdependence relationships refer to how the agents complement/combine each other's competence. The experimental results were obtained through a user study, using questionnaires and logged objective metrics. Our research found that teams with opportunistic interdependence relationships were significantly affected by trust violations compared to the baseline condition. Furthermore, although not as significant as the effect of trust violation, they also experienced a significant trust recovery during the tasks. Finally, the results of analyzing both subjective and objective fluency metrics did not give any significant result that indicates the difference in the level of collaboration fluency between the two conditions.

De Kindertelefoon is a children's helpline aimed at providing (pre-)adolescents with a person to talk to for a variety of subjects such as bullying, sex, and abuse. These people who talk for De Kindertelefoon need proper training and guidance. Among the tools that De Kindertelefoon provides is the 5-phase model, a conversational model. Researchers set up a simulation of a virtual child and De Kindertelefoon to help teach the 5-phase model. However, simulation alone was not enough, and we wanted to see if a feedback system could improve the results, with immediate feedback being the focus.
Immediate feedback is when the user fills in an answer and the system immediately provides information to strengthen knowledge.
When applied to an academic environment, both immediate feedback and student engagement have been proven to be important for completing a task.

Using a literature review, we found that immediate feedback can be directly linked to educational techniques such as self-improvement, self-efficacy, and the Self-Determination Theory. Through a focus group, we also found that constructive immediate feedback is an important pillar of De Kindertelefoon.

The design was achieved by looking at the existing and limited 5-phase model as a graph and trying to find an optimal path through that graph.

After conducting a within-subjects study experiment with 34 participants, the results were inconclusive, with neither condition appearing to be more useful for the group of participants, nor either condition being better at teaching them the 5-phase model.
As the results were inconclusive, the data was explored more by looking at it as a between-subject study, which showed that the explanation sheet might perform better for knowledge.

The research shows the possible strength of feedback in a practical manner. From the results, the immediate feedback is neither more nor less resonant than a classical approach to teaching the 5-phase model.

Joint attention is the shared focus multiple people can have on the same object and it is subconsciously used by humans every day. The simple act of verbally or non-verbally pointing out an object to one another, is a form of joint attention. Its use facilitates human cooperation, such as when someone needs to hand over an object to another person. When robots display those same joint attention behaviours, it could be useful in improving human-robot interactions and perhaps variably so in various levels of jointness.
This research investigates the effects of these levels of jointness on the interactions between a human and a robot. This is done by investigating the effect on the person's task performance, the effect on their mental model of the robot, and the effect on their perception of the robot as its own entity with its own mind. To this end, this research defines four levels of joint attention and designs an experiment that makes use of these four levels of joint attention. To perform this experiment, a system capable of establishing joint attention has been developed. The system is divided into two parts; hardware and software tools & applications which have not been developed in this project but are used as is, and software which has been actively developed in this research.
In this experiment, participants played three guessing games with a robot. Task performance was measured by the amount of time and hints needed, as well as the accuracy of a participant. The participant's mental model and perception of the robot as its own entity with its own mind were measured using a questionnaire.
This research found no significant effect on the mental model towards the robot and the perception of whether the robot is its own entity with its own mind, although it is not concluded that such an effect cannot be found.
This research did find a significant effect on task performance; higher levels of joint attention lead to faster task completion with less hints needed. But interestingly, this did not necessarily lead to a higher accuracy.
The system introduced by this research was originally intended to help children with autism spectrum disorder learn joint attention skills, but due to ethical and time constraints, the system was tested with and the experiment was performed with adults instead. Due to ethical constraints, the participants in this research were not asked to disclose whether they had an autism spectrum disorder diagnosis or not. However, future research could replicate this study with the intended target group.

Physical activity is one of the main factors that contribute to reducing the chance of chronic diseases such as cardiovascular disease, obesity, and depression, all while improving an individual’s health in general. While this is the case, the fact still remains that many adults across the world do not reach the minimum recommendations for physical activity. Setting physical activity goals is one of the most common approaches found in both self-help fitness apps, or medical interventions for increasing physical activity. The issue is that goal setting on its own cannot help people become more physically active, since, if they are not committed to reaching a goal, a person will just abandon it instead. Recent literature shows that creating plans for when to perform physical activity has the potential to help people commit to reaching their goal. Being committed to following the plan is therefore important for ultimately reaching the goal, but only creating the plan is not enough, since people might abandon it. Furthermore, dyadic planning, in which a helper aids the person in creating the plan, has produced even better results than individual planning. Thus, the aim of this work is to develop a virtual coach, which plays the role of the helper in dyadic planning and motivates people to commit themselves to following the plan, so that they can reach their physical activity goals. To facilitate this process, the virtual coach, named Jamie, operates based on reinforcement learning, giving it the ability to select the best persuasive strategy to use. It does so by taking into account the person's situation (opinions of the plan and of planning in general), as well as how these opinions might change based on what persuasive strategy the agent chooses to employ. The persuasive strategies considered were: proposing to make changes to the plan, explaining why planning is useful, identifying and dealing with barriers, and showing testimonials from other people who created plans and used them to reach their goals. Through an observational study, data for the reinforcement learning model was gathered, and a model was trained on the data. Analysing the data revealed that the choice of persuasive strategy is not crucial, as all of them had similar effects on the person's situation, and these effects were small to moderate. At the same time, we saw moderate differences when comparing the situation at the beginning and end of the conversation, indicating that the combined effect of multiple persuasive strategies is needed to change a person's situation. We also investigated the effect of including or excluding the person's situation from the model. If the person's situation is disregarded, the resulting model is equivalent to following a fixed order of persuasive strategies and, through simulations, we have shown that it can change about 74% of people's situations into one where they are likely to commit to the plan. When the person's situation is included, the percentage rises to 82%, suggesting that the person's situation is also important to consider. Both of these models have advantages and disadvantages, which are discussed and addressed. Thus, this thesis provides two models for a virtual coach that can hold a persuasive dialogue in the context of dyadic planning for physical activity, which can be used as the basis for systems which target behaviour change.