The usage of automated decision making-systems by public institutions letting the system decide on the approval, determination or denial of individuals benefits as an example, is an effective measure in making more amount of work done in a shorter time period and to a lower cost than if it would have been done by humans. But still, although the technology has developed into being able to help us in this way, so has also the potential problems that these systems can cause while they are operating. The ones primarily affected here will be the individuals that are denied their benefits, health care, or pensions. The systems can maintain hidden, historical stigmatizations and prejudices, disproportionally affecting members of a certain historically marginalized group in a negative way through its decisions, simply because the systems have learned to do so. There is also a risk that the actual programmer includes her or his own bias, as well as incorrect translation of applicable legislations or policies causing the finalized system to make decisions on unknown bases, demanding more, less or completely other things than those requirements that are set up by the public and written laws. The language in which these systems works are in mathematical algorithms, which most ordinary individuals, public employees or courts will not understand. If suspecting that you could have been discriminated against by an automated decision, the requirements for successfully claim a violation of discrimination in US-, Canadian- and Swedish courts, ECtHR and ECJ demands you to show on which of your characteristics you were discriminated, and in comparison to which other group, a group that instead has been advantaged. Still, without any reasons or explanations to why the decision has been taken available for you as an applicant or for the court responsible, the inability to identify such comparator can lead to several cases of actual indirect discriminations being denied. A solution to this could be to follow the advice of Sophia Moreau’s theory, focusing on the actual harm that the individual claim to have suffered instead of on categorizing her or him due to certain traits, or on finding a suitable comparator. This is similar to a ruling of the Swedish Court of Appeal, where a comparator was not necessary in order to establish that the applicant had been indirectly discriminated by a public institution. Instead, the biggest focus in this case was on the harm that the applicant claimed to have suffered, and then on investigating whether this difference in treatment could be objectively justified. In order for Swedish and European legislation to be able to meet the challenges that can arise through the usage of automated decision making-systems, this model of the Swedish Court of Appeal could be a better suited model to help individuals being affected by an automated decision of a public institution, being potentially indirectly discriminative.

The increased use of automated systems in the public sector has led to two types of processes, fully automated decision making and humans making decisions assisted by automated decision support systems (ADSS). While having a human in the loop is often motivated by having them act as a “safeguard” for imperfect automated systems, humans themselves are not perfect decision makers. Automation bias, a tendency to agree with the recommendations of automated systems even when they are wrong, is one problem facing humans using ADSS. Mainly found in monitoring tasks such as autopilots, it has also been studied in clinical decision support systems. The aim of this systematic review is to investigate whether automation bias poses a risk for ADSS in the public sector, and to identify possible moderators. Thirteen studies were included. By doing a narrative synthesis of the included studies I found mixed results for the existence of automation bias. While there is a lack of strong evidence for automation bias, even low levels could result in consequences for the public sector as these are decisions that impact citizens' everyday life. A number of moderators are identified and suggestions for system designers are made. / En ökning av automatiserade system inom offentlig sektor har lett till två typer av processer, helt automatiserade beslut och människor som tar beslut stödda av automatiserade beslutsstöd. Att ha en människa i processen är ofta motiverat av att använda dem som ett skydd mot bristfälliga automatiserade system, men människor är i sig själva inte perfekta beslutstagare. Automation bias, en tendens att följa rekommendationer från automatiserade system även när de är inkorrekta, är ett problem för människor som använder automatiserade system. Det har främst studerats i autopiloter, men också i kliniska beslutsstöd. Syftet med denna systematiska litteraturöversikt var att undersöka om automation bias är en risk för automatiserade beslutsstöd i offentlig sektor, och att identifiera möjliga moderatorer. Tretton studier inkluderades. Genom att genomföra en narrativ syntes fann jag blandade slutsatser gällande automation bias. Samtidigt som det finns begränsade starka bevis för automation bias, kan även de nivåerna resultera i konsekvenser för offentlig sektor då de tar beslut som påverkar befolkningens vardag. Ett antal moderatorer identifierades och förslag till systemdesigners presenteras.

The thesis investigates the feasibility of automated decision-making (ADM) in project management from two perspectives - technical feasibility, analysed through a comprehensive literature review, and organisational acceptance, evaluated through empirical evidence. To address technical feasibility, the literature study is used, and it underscores the significance of data-driven decision-making and the impact of advancements in machine learning. Organisational acceptance is investigated with thematic analysis, and the complementary method employed is semi-structured interviews, allowing for in-depth insights from experienced project managers. The analysis reveals that project integration, cost, and risk management exhibit considerable potential for ADM integration, whereas schedule, resource, and procurement management demonstrate varying levels of applicability. In contrast, scope, quality, communication, and stakeholder management are deemed less feasible due to their complex nature and the critical involvement of skilful project managers. As a result, the study advocates a balanced approach to ADM implementation, combining automated capabilities with human expertise. Its contributions lie in the formalisation and categorisation of ADM applications, addressing the challenges, and providing valuable insights for practical ADM adoption in project management.

Redes probabilísticas são modelos muito versáteis, com aplicabilidade crescente em diversas áreas. Esses modelos são capazes de estruturar e mensurar a interação entre variáveis, permitindo que sejam realizados vários tipos de análises, desde diagnósticos de causas para algum fenômeno até previsões sobre algum evento, além de permitirem a construção de modelos de tomadas de decisões automatizadas. Neste trabalho são apresentadas as etapas para a construção dessas redes e alguns métodos usados para tal, dando maior ênfase para as chamadas redes bayesianas, uma subclasse de modelos de redes probabilísticas. A modelagem de uma rede bayesiana pode ser dividida em três etapas: seleção de variáveis, construção da estrutura da rede e estimação de probabilidades. A etapa de seleção de variáveis é usualmente feita com base nos conhecimentos subjetivos sobre o assunto estudado. A construção da estrutura pode ser realizada manualmente, levando em conta relações de causalidade entre as variáveis selecionadas, ou semi-automaticamente, através do uso de algoritmos. A última etapa, de estimação de probabilidades, pode ser feita seguindo duas abordagens principais: uma frequentista, em que os parâmetros são considerados fixos, e outra bayesiana, na qual os parâmetros são tratados como variáveis aleatórias. Além da teoria contida no trabalho, mostrando as relações entre a teoria de grafos e a construção probabilística das redes, também são apresentadas algumas aplicações desses modelos, dando destaque a problemas nas áreas de marketing e finanças. / Probabilistic networks are very versatile models, with growing applicability in many areas. These models are capable of structuring and measuring the interaction among variables, making possible various types of analyses, such as diagnoses of causes for a phenomenon and predictions about some event, besides allowing the construction of automated decision-making models. This work presents the necessary steps to construct those networks and methods used to doing so, emphasizing the so called Bayesian networks, a subclass of probabilistic networks. The Bayesian network modeling is divided in three steps: variables selection, structure learning and estimation of probabilities. The variables selection step is usually based on subjective knowledge about the studied topic. The structure learning can be performed manually, taking into account the causal relations among variables, or semi-automatically, through the use of algorithms. The last step, of probabilities estimation, can be treated following two main approaches: by the frequentist approach, where parameters are considered fixed, and by the Bayesian approach, in which parameters are treated as random variables. Besides the theory contained in this work, showing the relations between graph theory and the construction of probabilistic networks, applications of these models are presented, highlighting problems in marketing and finance.

Redes probabilísticas são modelos muito versáteis, com aplicabilidade crescente em diversas áreas. Esses modelos são capazes de estruturar e mensurar a interação entre variáveis, permitindo que sejam realizados vários tipos de análises, desde diagnósticos de causas para algum fenômeno até previsões sobre algum evento, além de permitirem a construção de modelos de tomadas de decisões automatizadas. Neste trabalho são apresentadas as etapas para a construção dessas redes e alguns métodos usados para tal, dando maior ênfase para as chamadas redes bayesianas, uma subclasse de modelos de redes probabilísticas. A modelagem de uma rede bayesiana pode ser dividida em três etapas: seleção de variáveis, construção da estrutura da rede e estimação de probabilidades. A etapa de seleção de variáveis é usualmente feita com base nos conhecimentos subjetivos sobre o assunto estudado. A construção da estrutura pode ser realizada manualmente, levando em conta relações de causalidade entre as variáveis selecionadas, ou semi-automaticamente, através do uso de algoritmos. A última etapa, de estimação de probabilidades, pode ser feita seguindo duas abordagens principais: uma frequentista, em que os parâmetros são considerados fixos, e outra bayesiana, na qual os parâmetros são tratados como variáveis aleatórias. Além da teoria contida no trabalho, mostrando as relações entre a teoria de grafos e a construção probabilística das redes, também são apresentadas algumas aplicações desses modelos, dando destaque a problemas nas áreas de marketing e finanças. / Probabilistic networks are very versatile models, with growing applicability in many areas. These models are capable of structuring and measuring the interaction among variables, making possible various types of analyses, such as diagnoses of causes for a phenomenon and predictions about some event, besides allowing the construction of automated decision-making models. This work presents the necessary steps to construct those networks and methods used to doing so, emphasizing the so called Bayesian networks, a subclass of probabilistic networks. The Bayesian network modeling is divided in three steps: variables selection, structure learning and estimation of probabilities. The variables selection step is usually based on subjective knowledge about the studied topic. The structure learning can be performed manually, taking into account the causal relations among variables, or semi-automatically, through the use of algorithms. The last step, of probabilities estimation, can be treated following two main approaches: by the frequentist approach, where parameters are considered fixed, and by the Bayesian approach, in which parameters are treated as random variables. Besides the theory contained in this work, showing the relations between graph theory and the construction of probabilistic networks, applications of these models are presented, highlighting problems in marketing and finance.

The automation of processes is rapidly expanding across various industries, driven by technological advancements and the increasing reliance on data-driven solutions. The automation of processes can benefit all stakeholders and adversely affect fundamental human rights, such as the right to privacy and data protection. The risks to fundamental human rights are addressed by the GDPR, primarily by Article 22. This article set out the right of an individual not to be subject to a decision based solely on automated processing, including profiling, which produces legal effects concerning an individual or similarly significantly affects them.  The CJEU was asked to interpret Article 22 GDPR in ‘SCHUFA case’. The dispute arose from a credit score produced by SCHUFA AG, which led a financial institution to refuse a loan to the individual. The CJEU held that producing credit scores is an ‘automated individual decision’ if a third party draws strongly on that credit score. Such an ‘automated individual decision’ is prohibited in principle by Article 22 GDPR, unless exceptions apply.  This research aims to understand how SCHUFA case has changed the legal framework for agencies and how they may adjust their business practices to ensure compliance with Article 22 GDPR while still ensure their competitive advantage.  The findings indicate that credit scoring is crucially embedded in agencies’ processes as it massively benefits their operations. Agencies engaged in ADM may adjust their practices to legally comply with Article 22 GDPR, however not without momentarily jeopardizing their competitive advantage. Strategic changes necessary to tackle recent legal uncertainty are accompanied by further cost and longer operational response timers.

Workplace studies in Human-Computer Interaction (HCI) is a research field that has expanded in an explosive way during the recent years. Today there is a wide range of theoretical approaches and methods to choose from, which makes it problematic to make methodological choices both in research and system design. While there have been several studies that assess the different approaches to workplace studies, there seems to be a lack of studies that explore the theoretical and methodological differences between more structured methods within the research field. In this thesis, a comparative theoretical and empirical analysis of three methods for workplace studies is being conducted to deal with the following research problem: What level of theoretical depth and methodological structure is appropriate when conducting methods for workplace studies to inform design of complex socio-technical systems? When using the two criterions descriptive power and application power, to assess Contextual Design (CD), Determining Information Flow Breakdown (DIB), and Capturing Semi-Automated Decision-Making (CASADEMA), important lessons are learned about which methods are acceptable and useful when the purpose is to inform system design.

Decisions have been automated since the early days. Ever since the rise of AI, ML and DataAnalytics, algorithmic decision-making has experienced a boom time. Nowadays, using AI withina company is said to be critical to the success of a company. Considering the point that it can bequite costly to develop AI/ ML and integrating it into decision-making, it is striking how littleresearch was put into the identification and analysis of its cost drivers by now. This thesis is acontribution to raise and the awareness of possible cost drivers to algorithmic decisions. Thetopic was divided in two subgroups. That is solely algorithms and hybrid decision-making. Asystematic literature review was conducted to create a theoretical base for further research. Thecost drivers for algorithms to make decisions without human interaction, the identified costdrivers identified can be found at Data Storage (including initial, floor rent, energy, service,disposal, and environmental costs), Data Processing, Transferring and Migrating. Additionally,social costs and the ones related to fairness as well as the ones related to algorithms themselves(Implementation and Design, Execution and Maintenance) could be found. Business Intelligenceused for decision making raises costs in Data quality, Update delays of cloud systems, Personneland Personnel training, Hardware, Software, Maintenance and Data Storage. Moreover, it isimportant to say that the recurrence of some costs was detected. Further research should go inthe direction of applicability of the theoretical costs in practice.

No description available.

Background: Advancements in machine learning (ML) techniques and voice technology offer the potential to harness voice as a new tool for developing decision-support tools in healthcare for the benefit of both healthcare providers and patients. Motivated by technological breakthroughs and the increasing integration of Artificial Intelligence (AI) and Machine Learning (ML) in healthcare, numerous studies aim to investigate the diagnostic potential of ML algorithms in the context of voice-affecting disorders. This thesis focuses on respiratory diseases such as Chronic Obstructive Pulmonary Disease (COPD) and explores the potential of a decision support tool that utilizes voice and ML. This exploration exemplifies the intricate relationship between voice and overall health through the lens of applied health technology (AHT. This interdisciplinary nature of research recognizes the need for accurate and efficient diagnostic tools. Objective: The objectives of this licentiate thesis are twofold. Firstly, a Systematic Literature Review (SLR) thoroughly investigates the current state of ML algorithms in detecting voice-affecting disorders, pinpointing existing gaps and suggesting directions for future research. Secondly, the study focuses on respiratory health, specifically COPD, employing ML techniques with a distinct emphasis on the vowel "A". The aim is to explore hidden information that could potentially be utilized for the binary classification of COPD vs no COPD. The creation of a new Swedish COPD voice classification dataset is anticipated to enhance the experimental and exploratory dimensions of the research. Methods: In order to have a holistic view of a research field, one of the commonly utilized methods is to scan and analyze the literature. Therefore, Paper I followed the methodology of an SLR where existing journal publications were scanned and synthesized to create a holistic view in the realm of ML techniques employed to experiment on voice-affecting disorders. Based on the results from the SLR, Paper II focused on the data collection and experimentation for the binary classification of COPD, which was one of the gaps identified in the first study. Three distinct ML algorithms were investigated on the collected datasets through voice features, which consisted of recordings collected through a mobile application from participants 18 years old and above, and the most utilized performance measures were computed for the best outcome.  Results: The summary of findings from Paper I reveals the dominance of Support Vector Machine (SVM) classifiers in voice disorder research, with Parkinson's Disease and Alzheimer's Disease as the most studied disorders. Gaps in research include underrepresented disorders, limited datasets in terms of number of participants, and a lack of interest in longitudinal studies. Paper II demonstrates promising results in COPD classification using ML and a newly developed dataset, offering insights into potential decision support tools for COPD diagnosis. Conclusion: The studies covered in this dissertation provide a comprehensive literature summary of ML techniques used to support decision-making on voice-affecting disorders for clinical outcomes. The findings contribute to understanding the diagnostic potential of using ML on vocal features and highlight avenues for future research and technology development. Nonetheless, the experiment reveals the potential of employing voice as a digital biomarker for COPD diagnosis using ML.