This case study explored perceived artificial intelligence (AI) performance and intended future use (IFU) in users of AI-based applications. Users could become less motivated to use these applications if AIs do not clearly communicate their actions. A prototype, a user test, and a structured interview were iteratively developed. Eight students participated in the final iteration, which was thematically analyzed. The results indicate that an AI-based application that shows recommendations can positively affect perceived AI performance and IFU. Possibly, the recommendations increased users’ understanding of AI decisions, as well as their satisfaction. Therefore, recommendations could be a potential design element for increasing perceived AI performance and IFU. Finally, time-saving functionality is a design element that could lead to higher IFU in AI-based applications, possibly only for other tasks than examining recommendations. Further research needs to test these findings under different circumstances.

GranCloud is new application for the generation of real-time granular synthesis in the SuperCollider programming environment. Although the software was initially programmed for use in the interactive composition Creo, it was implemented as an independent program for use in any computer music project. GranCloud consists of a set of SuperCollider classes representing granular clouds and parameter objects defining control data for the synthesis. The software is very flexible, allowing users to create their own grain synthesis definitions and control parameters. Cloud objects encapsulate all of the control data and methods necessary to render virtually any type of granular synthesis. Parameter objects provide several simple methods for mapping grain parameters to complex changing data sets or to external data sources. GranCloud simplifies the complex task of generating granular synthesis, allowing composers to focus less on technological issues and more on musical considerations during the compositional process.

FAD * for Stadsgårdskajen Re-introduction of some of the knowledge from the master builder to the contemporary architect. This project investigates the interaction between designing architecture and building architecture in a computer controlled context. Algorithms have been developed to simulate and control an automated in situ brick stacking process. Brick laying is a well researched topic for robotic processes. It is an ideal material for building with robots (in Stockholm) since it is a generic building block that is cheap and easy to produce locally. Corbeled vaults, instead of keystone vaults, can be built without support and with all identical elements and are therefore also suitable for robotic fabrication.  The algorithms have been incorporated in a parametric model that have been used to develop public building at Stadsgårdskajen, Stockholm. FAD is about creating and explore a different architectural expression as well as reintroducing the brick as a relevant construction material for Stockholm. / FAD* för Stadsgårdskajen Återinförande av några av byggmästarens förmågor till den samtida arkitekten I detta examensarbete undersöks förhållandet mellan att designa arkitektur och att bygga arkitektur i en datorkontrollerad miljö. Algoritmer har utarbetats för att simulera och styra en automatiserad in situ murningsprocess. Inom fabrikation och automatisering med industrirobotar är tegelband ett välundersökt ämne. Tegel är ett idealiskt material vid byggande med robotar. Det är enkelt och billigt att framställa och tegel som byggkloss är en lämplig typologi för addativ fabrikation med robotar. Valv byggda genom utkragning istället för traditionella valvbågar kan byggas utan stöd och med identiska element. Därför är denna byggnadsteknik även lämplig för additiv fabrikation med robotoar. Algoritmerna har intergrerats i en parametrisk modell som har använts för att formge en offentlig byggnad på Stadsgårdskajen i Stockholm. FAD handlar såväl om skapande och utforskande av ett nytt arkitekturspråk som av återinförande av tegel som ett betydelsefullt konstruktionsmaterial i Stockholm.

Introduction: This qualitative case study extends the research of algorithmic management by examining the experiences of food-delivery workers working for the gig company Foodora. As Foodora promotes the job as flexible and autonomous, the utilisation of labour process theory (LPT) in this study helps to uncover if these sentiments hold true in practice by examining what Foodora’s labour process looks like, and how control, autonomy, and individualism take shape in the workplace. Method: Interviews with eight Foodora delivery workers working in Sweden were carried out, where the data was transcribed and coded. Analysis: Using LPT as a framework for the analysis, the material was coded according to the themes of control, transparency, resistance, consent, and individualism. By doing this, the study contextualises algorithmic management within the wider framework of capitalistic management forms and highlights how it impacts worker experiences. Results: The results show that the experiences of the job were largely different among the individual participants of the study. It is proved that Foodora’s employment of algorithmic management impacts almost every aspect of the daily work since it centres around following automated directives which the workers receive through an app. Individualism is accentuated by a self-entrepreneurial discourse promoted by Foodora and the fact that the workforce is dispersed with limited opportunities to interact. Despite being a dispersed workforce, the workers have been able to organise which has led to a growing number of workers joining the union. Conclusion: The impacts of algorithmic management are proved to be evident. The varied experiences of the job could have multiple explanations and needs to be explored further in relation to economical and societal factors. The effects of the newly implemented collective bargaining agreement also need to be examined in future research.

A plethora of high-impact applications involve predictive modeling of structured data. In various domains, from hospital readmission prediction in the medical realm, though weather forecasting and event detection in power systems, up to conversion prediction in online businesses, the data holds a certain underlying structure. Building predictive models from such data calls for leveraging the structure as an additional source of information. Thus, a broad range of structure-aware approaches have been introduced, yet certain common challenges in many structured learning scenarios remain unresolved. This dissertation revolves around addressing the challenges of scalability, algorithmic stability and temporal awareness in several scenarios of learning from either graphically or sequentially structured data. Initially, the first two challenges are discussed from a structured regression standpoint. The studies addressing these challenges aim at designing scalable and algorithmically stable models for structured data, without compromising their prediction performance. It is further inspected whether such models can be applied to both static and dynamic (time-varying) graph data. To that end, a structured ensemble model is proposed to scale with the size of temporal graphs, while making stable and reliable yet accurate predictions on a real-world application involving gene expression prediction. In the case of static graphs, a theoretical insight is provided on the relationship between algorithmic stability and generalization in a structured regression setting. A stability-based objective function is designed to indirectly control the stability of a collaborative ensemble regressor, yielding generalization performance improvements on structured regression applications as diverse as predicting housing prices based on real-estate transactions and readmission prediction from hospital records. Modeling data that holds a sequential rather than a graphical structure requires addressing temporal awareness as one of the major challenges. In that regard, a model is proposed to generate time-aware representations of user activity sequences, intended to be seamlessly applicable across different user-related tasks, while sidestepping the burden of task-driven feature engineering. The quality and effectiveness of the time-aware user representations led to predictive performance improvements over state-of-the-art models on multiple large-scale conversion prediction tasks. Sequential data is also analyzed from the perspective of a high-impact application in the realm of power systems. Namely, detecting and further classifying disturbance events, as an important aspect of risk mitigation in power systems, is typically centered on the challenges of capturing structural characteristics in sequential synchrophasor recordings. Therefore, a thorough comparative analysis was conducted by assessing various traditional as well as more sophisticated event classification models under different domain-expert-assisted labeling scenarios. The experimental findings provide evidence that hierarchical convolutional neural networks (HCNNs), capable of automatically learning time-invariant feature transformations that preserve the structural characteristics of the synchrophasor signals, consistently outperform traditional model variants. Their performance is observed to further improve as more data are inspected by a domain expert, while smaller fractions of solely expert-inspected signals are already sufficient for HCNNs to achieve satisfactory event classification accuracy. Finally, insights into the impact of the domain expertise on the downstream classification performance are also discussed. / Computer and Information Science

SkePU 3 is a framework for parallel program execution that uses higher order functions called skeletons, which provide a layer of abstraction between user code and the parallel implementation it provides through its backends. The backend that enables SkePU to run on an HPC cluster has a slowdown of a factor two. This reduces the viability of SkePU as an alternative for HPC, and as such, warrants an investigation. Programs written in SkePU are sequential-looking, single-source C++ programs where skeleton calls can transparently execute on multiple different types of processing units, such as CPU cores, GPUs and clusters, using different backends. In this thesis, a strategy for improving the performance of SkePU on clusters is presented, and with it, the design and implementation of a new cluster backend that is simpler and more closely integrated with the non-cluster SkePU code base. Runtime measurements are made, which show that the new cluster backend sees a relative speedup of about a factor of two, which effectively eliminates the slowdown.

TikTok differs from how the media audience is used to seeing other social media platforms. When opening the app, the user does not see the friends you have decided to follow. Instead, you are faced with an algorithmically decided and never-ending feed of content personified individually for the specific user. In recent years, TikTok has become a prominent and rapidly growing platform, especially among younger media users. The emergence of TikTok usage among the youth implies an increased need for examining cultural phenomena that are performed on this platform. This study will be exploring the emergence of the online phenomenon known as mediated public shaming on TikTok by critically reviewing how young Swedish TikTok users are engaging in and experiencing the mediated public shaming that is occurring on TikTok as well as its effects by taking a theoretical departure from media literacy theories. The study is a qualitative reception study where the empirical material is conducted by one-to-one interviews with nine mundane Swedish TikTok users between the ages of 16–27. It was discovered that young TikTok users have a complex relationship with the mediated public shaming they face on TikTok and engage in the concept in a rather objective and reflective manner. They do, however, express concern that the design of the media, in conjunction with mediated public shaming, may allow for subconscious effects of their opinions. Looking forward, there is an indicated compelling need for continued research within the field, proposedly further research of online cultures’ roles in the determination of platform design and construction of algorithms.

Technology is rapidly becoming more compact and engineers are investigating new applications that are possible with this compact technology. For example photography has recently become an obvious part in our lives due to the shrinking of cameras. If there was a way to implement a light spectrum sensor into our phones there are a lot of applications this could be used for. For example the visual light spectrum can be used to detect the difference between materials in ways that our eyes can not. It could also be used to detect the amount of UV-light when outdoors or be used as an infrared camera. In order for this to be possible the spectrum sensor can’t be too big which makes the common spectrometer a bad choice for the application. This thesis compares two methods of detecting visual light spectrum suitable for On-Chip applications. One method utilize well-defined optical filters while the other works by incorporating an algorithm that boosts the performance of less well-defined optical filters. A simulation was created to assess the performance of the methods and one of these spectrometers were then designed into a circuit board to test the performance in real environments. The results concluded that the algorithmic method currently has better performance to the same price but that the purely optically filtered method is set to be stronger in the future. / Dagens teknik blir snabbt mer kompakt och ingenjörer undersöker nya tillämpningar som är möjliga med den här kompakta tekniken. Till exempel fotografi har nyligen fått en självklar plats i våra liv när kameratekniken blivit mindre. Om det fanns ett sätt att implementera en ljusspektrumsensor i våra telefoner så hade det funnits en mängd applikationer för detta. Till exempel så kan ljusspektrumet användas för att märka skillnad på material som våra ögon inte kan. Det kan också användas för att mäta mängden UV-ljus när man är ute, eller användas som en infraröd kamera. För att det här ska vara möjligt så kan inte spektrumsensorn vara alltför stor vilket gör en traditionell spektrometer blir olämplig. Den här avhandlingen jämför två metoder för att mäta det synliga ljusspektrumet som är passande för att placeras direkt på ett kretskort. En metod använder specifika optiska filter medan den andra metoden använder en algoritm för att förbättra utslaget hos mindre specifika optiska filter. En simulation skapades för att bedöma de två metoderna och ett kretskort designades sedan för att implementera en av spektrometrarna på och testa beteendet i verkliga miljöer. Resultaten visade att den algoritmiska metoden just nu har bättre prestanda men att den rent optiskt filtrerade metoden troligtvis kommer vara starkare i framtiden.

Sättet människor konsumerar musik har förändrats genom historien och i samband med digitaliseringen har nya möjligheter att lyssna på musik uppstått. I samband med denna övergång har även fler algoritmbaserade tillvägagångssätt för musiklyssnande uppkommit. Syftet med studien är att undersöka unga vuxnas upplevelser av den algoritmbaserade Radiofunktionen på musikstreamingtjänsten Spotify. Studien har även undersökt huruvida de algoritmbaserade förslagen eventuellt påverkar användarnas upplevelser och musikbeteenden.Undersökningen är baserad på sju respondenters upplevelser inom åldersspannet 20-30 år, i denna studie definierat som unga vuxna. Studien har genomförts via kvalitativa metoder däribland en inledande dagboksstudie, med syfte att förbereda respondenternas reflektiva tänkande vilket följdes upp med semistrukturerade intervjuer. Därefter transkriberades intervjuerna för att senare kodas och en tematisk analys genomfördes. Resultatet av studien påvisar att det finns en variation i användandet av radiofunktionen samt att majoriteten av respondenterna uttryckt en positiv upplevelse av Radio på Spotify. Studien bidrar med nya insikter kring algoritmbaserade upplevelser hos användare i relation till musikstreaming och framförallt funktionen radio, samt hur användandet kan skilja sig beroende på situation och anledning till användande. / The streaming of music has during the last two decades become a new standard for how people acquire and listen to music. In correlation with this shift, other possibilities for listening to music have been on the rise. The purpose of this study is to investigate the experience for young adults of algorithm-based playlist ”Radio” on the streaming service platform Spotify. The investigation will determine whether or not these algorithm-based suggestions potentially affect the users' experience of listening to music and overall music behavior.The study is based on the experience of seven respondents of ages ranging between 20-30 years old - through this study referred to as "young adults." The qualitative methods this research has followed consists of a simple initial diary study, in order to prepare the respondents reflective thinking before the following semi-structured interviews. The interviews were then transcribed and followed by coding as well as a thematical analysis. The results of the study show that there is a variation in the use of the ”Radio” phenomenon on Spotify and that the vast majority of the respondents participating in the investigation expressed an overall positive experience. Furthermore, this study indicates that the respondents utilize the feature on different occasions and for different purposes.

The development of artificial intelligence (AI) has revolutionised the way we interact with technology and each other, both in society and in professional careers. Although they come with great potential for productivity and automation, AI systems have been found to exhibit biases that reflect and perpetuate existing societal inequalities. With the recent rise of artificial intelligence tools exploiting the large language model (LLM) technology, such as ChatGPT, Bing Chat and Bard AI, this research project aims to investigate the extent of AI bias in said tools and explore its ethical implications. By reviewing and analysing responses to carefully crafted prompts generated by three different AI chatbot tools, the author will intend to determine whether the content generated by these tools indeed exhibits patterns of bias related to various social identities, as well as compare the extent to which such bias is present across all three tools. This study will contribute to the growing body of literature on AI ethics and inform efforts to develop more equitable and inclusive AI systems. By exploring the ethical dimensions of AI bias in selected LLMs, this research will shed light on the broader societal implications of AI and the role of technology in shaping our future.