Denna kandidatuppsats undersöker implementeringen av artificiell intelligens (AI) och dess påverkan på den svenska spelutvecklingsbranschen. Genom kvalitativa intervjuer med branschprofessionella har vi kartlagt hur AI integreras i spelutvecklingsprocesser, påverkar yrkesroller och förändrar kompetensbehov. Studien belyser både möjligheter och utmaningar som AI medför, inklusive organisatoriska förändringar, ekonomiska fördelar och sociala aspekter. Resultaten visar att AI inte bara bidrar till teknisk innovation utan också kräver en strategisk anpassning av företagets arbetsprocesser och strukturer. Vår analys indikerar att en framgångsrik AI-implementering kan stärka företagens konkurrenskraft på en kompetitiv marknad. / This bachelor's thesis examines the implementation of artificial intelligence (AI) and its impact on the Swedish game development industry. Through qualitative interviews with industry professionals, we have mapped how AI is integrated into game development processes, affects job roles, and changes competency requirements. The study highlights both opportunities and challenges brought by AI, including organizational changes, economic benefits, and social aspects. The results show that AI not only contributes to technical innovation but also requires a strategic adjustment of companies work processes and structures. Our analysis indicates that successful AI implementation can enhance companies competitiveness in a competitive market.

Denna studie undersöker utmaningarna och möjligheterna med implementeringen av artificiell intelligens (AI) på arbetsplatsen. Genom en kombination av intervjuer, enkäter och litteraturöversikt analyseras olika aspekter av AI-implementering, inklusive effektivitet, produktivitet, kompetensutveckling, etiska dilemman och samarbete. Resultaten visar att AI-implementering kan leda till ökad effektivitet och produktivitet genom automatisering av rutinmässiga uppgifter. Dock finns det utmaningar relaterade till anställdas acceptans, utbildning och säkerhetsfrågor. Vidare betonas vikten av att etablera etiska riktlinjer för att hantera potentiella risker och dilemman som uppstår med AI-användning. Kommunikation och samarbete identifieras som nyckelfaktorer för en framgångsrik integration av AI på arbetsplatsen. Denna studie belyser behovet av att förstå och navigera de komplexa dynamikerna som omger AI-implementering för att maximera dess fördelar och minimera dess risker. / This study examines the challenges and opportunities associated with the implementation of artificial intelligence (AI) in the workplace. Through a combination of interviews, surveys, and literature review, various aspects of AI implementation are analyzed, including efficiency, productivity, skills development, ethical dilemmas, and collaboration. The results indicate that AI implementation can lead to increased efficiency and productivity by automating routine tasks. However, there are challenges related to employee acceptance, training, and security issues. Furthermore, the importance of establishing ethical guidelines to address potential risks and dilemmas arising from AI usage is emphasized. Communication and collaboration are identified as key factors for successful integration of AI in the workplace. This study highlights the need to understand and navigate the complex dynamics surrounding AI implementation to maximize its benefits and minimize its risks.

Submitted by Jos? Henrique Henrique (jose.neves@ufvjm.edu.br) on 2017-01-09T12:21:59Z No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) andre_vinicius_gon?alves.pdf: 1266538 bytes, checksum: 42c3fe90b9d66c8cb7b901a10e548f1b (MD5) / Approved for entry into archive by Rodrigo Martins Cruz (rodrigo.cruz@ufvjm.edu.br) on 2017-01-31T13:56:36Z (GMT) No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) andre_vinicius_gon?alves.pdf: 1266538 bytes, checksum: 42c3fe90b9d66c8cb7b901a10e548f1b (MD5) / Made available in DSpace on 2017-01-31T13:56:36Z (GMT). No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) andre_vinicius_gon?alves.pdf: 1266538 bytes, checksum: 42c3fe90b9d66c8cb7b901a10e548f1b (MD5) Previous issue date: 2016-06 / Nos ?ltimos dez anos muitos pesquisadores t?m realizado estudos sobre assist?ncia personalizada e inteligente em Ambientes Educacionais a Dist?ncia, baseada na identifica??o dos Estilos de Aprendizagem. Sabe-se que o aprendizado ? algo extremamente particular, pois cada estudante possui estilos pr?prios e pode sofrer mudan?as diante de situa??es diversas como, por exemplo, objetivo, motiva??o, personalidade, etc. Por isso, o conceito de adaptabilidade do conte?do did?tico tem se tornado de grande import?ncia na personaliza??o do Sistema de Gerenciamento de Aprendizagem (SGA). Diante desse fato, Dor?a (2012) prop?e uma abordagem de Sistema Adaptativo e Inteligente para Educa??o (SAIE), utilizando t?cnicas probabil?sticas e Intelig?ncia Artificial (IA), capaz de detectar e adaptar, de maneira din?mica e autom?tica, os estilos de aprendizagem do estudante, considerando o Modelo de Estilo de Aprendizagem Felder-Silverman?s. Ap?s pesquisa detalhada, foram propostas algumas adapta??es baseadas na abordagem original, alterando o funcionamento de dois componentes espec?ficos: o M?dulo Pedag?gico e o Componente de Modelagem do Estudante. Al?m disso, prop?e-se uma nova estrutura do Modelo Estudante, contemplando o hist?rico de desempenho do aluno nos processos avaliativos. Por conseguinte, realizaram-se testes para avaliar os impactos de tais mudan?as por meio uma compara??o estat?stica utilizando o m?todo T-Pareado. Pelos resultados obtidos, as ideias deste trabalho proporcionaram uma melhora m?dia de 6,07% no desempenho avaliativo do estudante e uma redu??o m?dia de 68,27% nos problemas de aprendizagem, demonstrando efici?ncia e efic?cia da proposta. / Disserta??o (Mestrado Profissional) ? Programa de P?s-Gradua??o em Educa??o, Universidade Federal dos Vales do Jequitinhonha e Mucuri, 2015. / Since last decade many researchers have been conducting studies on personalized and intelligent assistance in distance education based on identification of learning styles. It is known that learning is something very particular because each student has their own styles and are subject to change on a variety of situations such as goal, motivation, personality, etc. Therefore, this study discusses the concept of adaptability of educational content as a way to provide customization of Learning Management System (LMS). Through probabilistic techniques and Artificial Intelligence (AI), Dor?a (2012) proposed a approach Adaptive and Intelligent System for Education (AIES) able to dynamically and automatically detect, select and adapt learning objects based on the student?s profile through Felder-Silverman Learning Styles Model (FSLSM). After detailed study, it has been proposed some adaptations based on this approach, thereby altering the operation of two specific components: the Pedagogical Module and the Student Modeling Component. In addition, it is proposed a new structure Model Student, considering learner performance history in the evaluation processes. Therefore, it carried out tests to assess the impacts of such changes through a statistical comparison by T-Paired method. From the results, the ideas in this work provides an average improvement of 6.07% in the performance evaluation of the student and an average reduction of 68.27% in the learning problems, demonstrating proposal of efficiency and effectiveness.

It was analyzed if instructions on a head-worn Augmented Reality display (AR-HWD) are better for manual industrial assembly tasks than instructions on a stationary screen. A prototype was built which consisted of virtual instruction screens for two example assembly tasks. In a comparison study participants performed the tasks with instructions through an AR-HWD and alternatively through a stationary screen. Questionnaires, interviews and observation notes were used to evaluate the task performances and the user experience. The study revealed that the users were excited and enjoyed trying the technology. The perceived usefulness at the current state was diverse, but the users saw a huge potential in AR-HWDs for the future. The task accuracy with instructions on the AR-HWD was equally good as with instructions on the screen. AR-HWDs are found to be a better approach than a stationary screen, but technological limitations need to be overcome and workers need to train using the new technology to make its application efficient.

Denna uppsats presenterar resultaten av en kvalitativ undersökning som syftar till att ge en djupare förståelse för användningen av AI inom radiologi, dess framtida påverkan på yrket och hur det används idag. Genom att genomföra tre intervjuer med personer som arbetar inom radiologi, har datainsamlingen fokuserat på att identifiera de positiva och negativa aspekterna av AI i radiologi, samt dess potentiella konsekvenser på yrket. Resultaten visar på en allmän acceptans för AI inom radiologi och dess förmåga att förbättra diagnostiska processer och effektivisera arbetet. Samtidigt finns det en viss oro för att AI kan ersätta människor och minska behovet av mänskliga bedömningar. Denna uppsats ger en grundläggande förståelse för hur AI används inom radiologi och dess möjliga framtida konsekvenser. / This essay presents the results of a qualitative study aimed at gaining a deeper understanding of the use of artificial intelligence (AI) in radiology, its potential impact on the profession and how it’s used today. By conducting three interviews with individuals working in radiology, data collection focused on identifying the positive and negative aspects of AI in radiology, as well as its potential consequences on the profession. The results show a general acceptance of AI in radiology and its ability to improve diagnostic processes and streamline work. At the same time, there is a certain concern that AI may replace humans and reduce the need for human judgments. This report provides a basic understanding of how AI is used in radiology and its possible future consequences.

This project is about screw hole detection using neural networks for automated assembly and disassembly. In a lot of industrial companies, such as Ericsson AB, there are products such as radio units or filters that have a lot of screw holes. Thus, the assembly and disassemble process is very time consuming and demanding for a human to assemble and disassemble the products. The problem statement in this project is to investigate the performance of neural networks within object detection and semantic segmentation to detect screw holes in industrial products. Different industrial models were created and synthetic data was generated in Blender. Two types of experiments were done, the first one compared an object detection algorithm (Faster R-CNN) with a semantic segmentation algorithm (SegNet) to see which area is most suitable for hole detection. The results showed that semantic segmentation outperforms object detection when it comes to detect multiple small holes. The second experiment was to further investigate about semantic segmentation algorithms by adding U-Net, PSPNet and LinkNet into the comparison. The networks U-Net and LinkNet were the most successful ones and achieved a Mean Intersection over Union (MIoU) of around 0.9, which shows that they have potential for further development. Thus, conclusions draw in this project are that segmentation algorithms are more suitable for hole detection than object detection algorithms. Furthermore, it shows that there is potential in neural networks within semantic segmentation to detect screw holes because of the results of U-Net and LinkNet. Future work that one can do is to create more advanced product models, investigate other segmentation networks and hyperparameter tuning. / Det här projektet handlar om skruvhålsdetektering genom att använda neurala nätverk för automatiserad montering och demontering. I många industriföretag, såsom Ericsson AB, finns det många produkter som radioenheter eller filter som har många skruvhål. Därmed, är monterings - och demonteringsprocessen väldigt tidsfördröjande och krävande för en människa att montera och demontera produkterna. Problemformuleringen i detta projekt är att undersöka prestationen av olika neurala nätverk inom objekt detektering och semantisk segmentering för skurvhålsdetektering på indutriella produkter. Olika indutriella modeller var skapade och syntetisk data var genererat i Blender. Två typer av experiment gjordes, den första jämförde en objekt detekterings algoritm (Faster R-CNN) med en semantisk segmenterigs algoritm för att vilket område som är mest lämplig för hål detektering. Resultaten visade att semantisk segmentering utpresterar objekt detektering när det kommer till att detektera flera små hål. Det andra experimentet handlade om att vidare undersöka semantiska segmenterings algoritmer genom att addera U-Net, PSPNet och LinkNet till jämförelsen. Nätverken U-Net och PSPNet var de mest framgångsrika och uppnåde en Mean Intersection over Union (MIoU) på cirka 0.9, vilket visar på att de har potential för vidare utveckling. Slutsatserna inom detta projekt är att semantisk segmentering är mer lämplig för hål detektering än objekt detektering. Dessutom, visade sig att det finns potential i neurala nätverk inom semantisk segmentering för att detejtera skruvhål på grund av resultaten av U-Net och LinkNet. Framtida arbete som man kan göra är att skapa flera avancerade produkt modeller, undersöka andra segmenterisk nätverk och hyperparameter tuning.

The text-guided diffusion model GLIDE (Guided Language to Image Diffusion for Generation and Editing) is the state of the art in text-to-image generative artificial intelligence (AI). GLIDE has rich representations, but medical applications of this model have not been systematically explored. If GLIDE had useful medical knowledge, it could be used for medical image analysis tasks, a domain in which AI systems are still highly engineered towards a single use-case. Here we show that the publicly available GLIDE model has reasonably strong representations of key topics in cancer research and oncology, in particular the general style of histopathology images and multiple facets of diseases, pathological processes and laboratory assays. However, GLIDE seems to lack useful representations of the style and content of radiology data. Our findings demonstrate that domain-agnostic generative AI models can learn relevant medical concepts without explicit training. Thus, GLIDE and similar models might be useful for medical image processing tasks in the future - particularly with additional domain-specific fine-tuning.

SammanfattningI  denna  studie  utforskas  kapaciteterna  och  begränsningarna  hos  avancerade  stora språkmodeller (SSM), med särskilt fokus på ChatGPT-4 och Google Bard. Studien inleds med att ge en historisk bakgrund till artificiell intelligens och hur denna utveckling har lett fram till skapandet av dessa modeller. Därefter genomförs en kritisk analys av deras prestanda i språkbehandling och problemlösning. Genom att evaluera deras effektivitet i hanteringen av nyhetsinnehåll och sociala medier, samt i utförandet av kreativa uppgifter som pussel, belyses deras förmåga inom språklig bearbetning samt de utmaningar de möter i att förstå nyanser och utöva kreativt tänkande.I denna studie framkom det att SSM har en avancerad förmåga att förstå och reagera på komplexa språkstrukturer. Denna förmåga är dock inte utan begränsningar, speciellt när det kommer till uppgifter som kräver en noggrann bedömning för att skilja mellan sanning och osanning. Denna observation lyfter fram en kritisk aspekt av SSM:ernas nuvarande kapacitet, de är effektiva inom många områden, men möter fortfarande utmaningar i att hantera de finare nyanserna i mänskligt språk och tänkande. Studiens resultat betonar även vikten av mänsklig tillsyn vid användning av artificiell intelligens (AI), vilket pekar på behovet av att ha realistiska förväntningar på AI:s kapacitet och betonar vidare betydelsen av en ansvarsfull utveckling  av  AI,  där  en  noggrann  uppmärksamhet  kring etiska  aspekter  är  central.  En kombination av mänsklig intelligens och AI föreslås som en lösning för att hantera komplexa utmaningar, vilket bidrar till en fördjupad förståelse av avancerade språkmodellers dynamik och deras roll inom AI:s bredare utveckling och tillämpning.

L'évolution rapide des réseaux sociaux en ligne (RSO) représente un défi significatif dans l'identification et l'atténuation des fausses informations, incluant les fausses nouvelles, la désinformation et la mésinformation. Cette complexité est amplifiée dans les environnements numériques où les informations sont rapidement diffusées, nécessitant des stratégies sophistiquées pour différencier le contenu authentique du faux. L'un des principaux défis dans la détection automatique de fausses informations est leur présentation réaliste, ressemblant souvent de près aux faits vérifiables. Cela pose de considérables défis aux systèmes d'intelligence artificielle (IA), nécessitant des données supplémentaires de sources externes, telles que des vérifications par des tiers, pour discerner efficacement la vérité. Par conséquent, il y a une évolution technologique continue pour contrer la sophistication croissante des fausses informations, mettant au défi et avançant les capacités de l'IA. En réponse à ces défis, ma thèse introduit le cadre FACTS-ON (Fighting Against Counterfeit Truths in Online Social Networks), une approche complète et systématique pour combattre la désinformation dans les RSO. FACTS-ON intègre une série de systèmes avancés, chacun s'appuyant sur les capacités de son prédécesseur pour améliorer la stratégie globale de détection et d'atténuation des fausses informations. Je commence par présenter le cadre FACTS-ON, qui pose les fondements de ma solution, puis je détaille chaque système au sein du cadre : EXMULF (Explainable Multimodal Content-based Fake News Detection) se concentre sur l'analyse du texte et des images dans les contenus en ligne en utilisant des techniques multimodales avancées, couplées à une IA explicable pour fournir des évaluations transparentes et compréhensibles des fausses informations. En s'appuyant sur les bases d'EXMULF, MythXpose (Multimodal Content and Social Context-based System for Explainable False Information Detection with Personality Prediction) ajoute une couche d'analyse du contexte social en prédisant les traits de personnalité des utilisateurs des RSO, améliorant la détection et les stratégies d'intervention précoce contre la désinformation. ExFake (Explainable False Information Detection Based on Content, Context, and External Evidence) élargit encore le cadre, combinant l'analyse de contenu avec des insights du contexte social et des preuves externes. Il tire parti des données d'organisations de vérification des faits réputées et de comptes officiels, garantissant une approche plus complète et fiable de la détection de la désinformation. La méthodologie sophistiquée d'ExFake évalue non seulement le contenu des publications en ligne, mais prend également en compte le contexte plus large et corrobore les informations avec des sources externes crédibles, offrant ainsi une solution bien arrondie et robuste pour combattre les fausses informations dans les réseaux sociaux en ligne. Complétant le cadre, AFCC (Automated Fact-checkers Consensus and Credibility) traite l'hétérogénéité des évaluations des différentes organisations de vérification des faits. Il standardise ces évaluations et évalue la crédibilité des sources, fournissant une évaluation unifiée et fiable de l'information. Chaque système au sein du cadre FACTS-ON est rigoureusement évalué pour démontrer son efficacité dans la lutte contre la désinformation sur les RSO. Cette thèse détaille le développement, la mise en œuvre et l'évaluation complète de ces systèmes, soulignant leur contribution collective au domaine de la détection des fausses informations. La recherche ne met pas seulement en évidence les capacités actuelles dans la lutte contre la désinformation, mais prépare également le terrain pour de futures avancées dans ce domaine critique d'étude. / The rapid evolution of online social networks (OSN) presents a significant challenge in identifying and mitigating false information, which includes Fake News, Disinformation, and Misinformation. This complexity is amplified in digital environments where information is quickly disseminated, requiring sophisticated strategies to differentiate between genuine and false content. One of the primary challenges in automatically detecting false information is its realistic presentation, often closely resembling verifiable facts. This poses considerable challenges for artificial intelligence (AI) systems, necessitating additional data from external sources, such as third-party verifications, to effectively discern the truth. Consequently, there is a continuous technological evolution to counter the growing sophistication of false information, challenging and advancing the capabilities of AI. In response to these challenges, my dissertation introduces the FACTS-ON framework (Fighting Against Counterfeit Truths in Online Social Networks), a comprehensive and systematic approach to combat false information in OSNs. FACTS-ON integrates a series of advanced systems, each building upon the capabilities of its predecessor to enhance the overall strategy for detecting and mitigating false information. I begin by introducing the FACTS-ON framework, which sets the foundation for my solution, and then detail each system within the framework: EXMULF (Explainable Multimodal Content-based Fake News Detection) focuses on analyzing both text and image in online content using advanced multimodal techniques, coupled with explainable AI to provide transparent and understandable assessments of false information. Building upon EXMULF’s foundation, MythXpose (Multimodal Content and Social Context-based System for Explainable False Information Detection with Personality Prediction) adds a layer of social context analysis by predicting the personality traits of OSN users, enhancing the detection and early intervention strategies against false information. ExFake (Explainable False Information Detection Based on Content, Context, and External Evidence) further expands the framework, combining content analysis with insights from social context and external evidence. It leverages data from reputable fact-checking organizations and official social accounts, ensuring a more comprehensive and reliable approach to the detection of false information. ExFake's sophisticated methodology not only evaluates the content of online posts but also considers the broader context and corroborates information with external, credible sources, thereby offering a well-rounded and robust solution for combating false information in online social networks. Completing the framework, AFCC (Automated Fact-checkers Consensus and Credibility) addresses the heterogeneity of ratings from various fact-checking organizations. It standardizes these ratings and assesses the credibility of the sources, providing a unified and trustworthy assessment of information. Each system within the FACTS-ON framework is rigorously evaluated to demonstrate its effectiveness in combating false information on OSN. This dissertation details the development, implementation, and comprehensive evaluation of these systems, highlighting their collective contribution to the field of false information detection. The research not only showcases the current capabilities in addressing false information but also sets the stage for future advancements in this critical area of study.

In the rapidly advancing field of artificial intelligence, Large Language Models (LLMs) like GPT-3, GPT-4, and Gemini have revolutionized sectors by automating complex tasks. Despite their advancements, LLMs and more noticeably smaller language models (SLMs) still face challenges, such as generating unfounded content "hallucinations." This project aims to enhance SLMs for broader accessibility without extensive computational infrastructure. By supervised fine-tuning of smaller models with new datasets, SQUAD-ei and SQUAD-GPT, the resulting model, KERMIT-7B, achieved superior performance in TYDIQA-GoldP, demonstrating improved information extraction while retaining generative quality. / Inom det snabbt växande området artificiell intelligens har stora språkmodeller (LLM) som GPT-3, GPT-4 och Gemini revolutionerat sektorer genom att automatisera komplexa uppgifter. Trots sina framsteg stårdessa modeller, framför allt mindre språkmodeller (SLMs) fortfarande inför utmaningar, till exempel attgenerera ogrundat innehåll "hallucinationer". Denna studie syftar till att förbättra SLMs för bredare till-gänglighet utan krävande infrastruktur. Genom supervised fine-tuning av mindre modeller med nya data-set, SQUAD-ei och SQUAD-GPT, uppnådde den resulterande modellen, KERMIT-7B, överlägsen pre-standa i TYDIQA-GoldP, vilket visar förbättrad informationsutvinning samtidigt som den generativa kva-liteten bibehålls.