AI Research Research Articles

Pluralising critical technical practice

In this special issue, we turn to ideas of and approaches to critical technical practices (CTPs) as entry points to doing critique and doing things critically in digitally mediated cultures and societies. We explore the pluralisation of ‘critical technical practice’, starting from its early formulations in the context of AI research and development (Agre, 1997a, 1997b) to the many ways in which it has resonated and been taken up by different publications, projects, groups, and communities of practice, and what is has come to mean. Agre defined CTP as a situational, practical, and constructive way of working: ‘a technical practice for which critical reflection upon the practice is part of the practice itself’ (1997a: XII). Communities of practice in which the notion has been adopted, adapted, and put to use range from human–computer interaction (HCI) to media art and pedagogy, from science and technology studies (STS) and computer-supported cooperative work (CSCW) to digital humanities, media studies and data studies. This special issue affirms the pluralisation of CTP, and serves as an invitation to (re)consider what it means to use this notion drawing on a wider body of work, including beyond Agre. In this introduction, we review and discuss CTPs according to (1) Agre, (2) indexed research, and (3) contributors to this special issue. We conclude with some questions and considerations for those interested in working with this notion.

LIMEADE: From AI Explanations to Advice Taking

Research in human-centered AI has shown the benefits of systems that can explain their predictions. Methods that allow AI to take advice from humans in response to explanations are similarly useful. While both capabilities are well developed for transparent learning models (e.g., linear models and GA 2 Ms) and recent techniques (e.g., LIME and SHAP) can generate explanations for opaque models, little attention has been given to advice methods for opaque models. This article introduces LIMEADE, the first general framework that translates both positive and negative advice (expressed using high-level vocabulary such as that employed by post hoc explanations) into an update to an arbitrary, underlying opaque model. We demonstrate the generality of our approach with case studies on 70 real-world models across two broad domains: image classification and text recommendation. We show that our method improves accuracy compared to a rigorous baseline on the image classification domains. For the text modality, we apply our framework to a neural recommender system for scientific papers on a public website; our user study shows that our framework leads to significantly higher perceived user control, trust, and satisfaction.

Facilitating the implementation of AI-based assistive technologies for persons with disabilities in vocational rehabilitation: A practical design thinking approach

What is vocational learning? It refers to education and training that focuses on acquiring practical skills and knowledge directly related to specific jobs or careers. All over the world, it is unfortunate that a person with a disability is less likely to be employed than one without. AI has served as an important tool in providing assistive technologies for persons with disabilities in vocational learning. Luckily, design thinking which is an extension of innovation that allows you to design solutions for end users with a single problem statement in mind was developed.
 In this research we sought out ways in which AI has helped those with disabilities. Are the AIs really effective? Do they come with any challenges? Can these problems be solved? We discovered some challenges encountered through the use of these AI and found a way to solve them. By leveraging AI algorithms, researchers and developers can create cost-effective solutions that cater to a wide range of disabilities.
 This includes AI-powered prosthetics, smart glasses for visual assistance, AI voice technologies, such as Siri, Alexa, accessibility tools such as screen readers, captioning services and other assistive devices that enhance mobility, communication, and overall independence and autonomy. AI can track and analyse an individual’s progress in vocational training then adaptive learning platforms can adjust the difficulty and pace of training materials based on the learner’s performance and progress.
 Incorporating AI into vocational training can significantly enhance accessibility, individualised support, and the overall learning experience for people with disabilities. It empowers them to acquire the skills and knowledge they need to pursue meaningful careers and contributes to greater inclusivity in the workforce.

The Priesthood of All Sentients? The Eclipse of Transcendence in an Age of Artificial Intelligence

BlessU-2 is a robot and a specifically designed anthropomorphic machine that performed a blessing service during 2017’s exhibition in Wittenberg, Germany, commemorating 500 years of the reformation. The exhibition was intended to open a conversation about the future of the church’s service and AI’s role in it. This research explores and introduces today’s AI research development and theological challenges for ecclesiology and Christian worship. The research explores the theological themes that can be employed by Christian theology in general and ecclesiology in particular. It proposes that the main challenge of AI developments is the reinterpretation of the transcendent in the form of AI in which technology is perceived as godlike. On the other hand, AI fills a void in the religious lacuna that reaffirms the yearning for meaning and connection. Finally, the paper proposes the importance of the basic human nature of relationality in an age of simulated artificial reality.

Experiences of using artificial intelligence in healthcare: a qualitative study of UK clinician and key stakeholder perspectives

ObjectivesArtificial intelligence (AI) is a rapidly developing field in healthcare, with tools being developed across various specialties to support healthcare professionals and reduce workloads. It is important to understand the...

Face Authentication using MTCNN and FaceNet

Abstract: Advancements in facial recognition are on the rise after the recent breakthroughs in deep learning technologies and the extensive training datasets. But still, the practical application of facial recognition for authentication purposes faces some difficulties and problems when dealing with variations that are observed in real-world scenarios, including variations in facial poses, varying angles, lightning conditions and potential obstructions. In order to resolve these challenges, this paper presents an innovative deep learning API, which combines the capabilities of MTCNN and FaceNet, to overcome these limitations. Building upon the foundations of MTCNN and FaceNet, we have developed an API capable of verifying the identity of users through a two-step verification process. The system utilizes MTCNN for precise face detection and FaceNet for facial feature extraction that has the ability to directly map facial images into a compact Euclidean space in which spatial distances directly reflect facial similarity, and then storing this data securely within a MongoDB database. We have used the cutting-edge Deep Face model, developed by Facebook’s AI research team. Its reliable deep learning capabilities have helped us in achieving exceptional accuracy in the user’s verification process. This system serves to provides a superior level of user authentication, effectively reducing the risk of an unauthorized access while optimizing the user experience. Even though there is still some room for improvement within our experiment, we are committed to providing a comprehensive summary of our findings and an exploration of the challenges on the horizon.

“Machine Down”: making sense of human–computer interaction—Garfinkel’s research on ELIZA and LYRIC from 1967 to 1969 and its contemporary relevance

AbstractThis paper examines Harold Garfinkel’s work with ELIZA and a related program LYRIC from 1967 to 1969. AI researchers have tended to treat successful human–machine interaction as if it relied primarily on non-human machine characteristics, and thus the often-reported attribution of human-like qualities to communication with computers has been criticized as a misperception—and humans who make such reports referred to as “deluded.” By contrast Garfinkel, building on two decades of prior research on information and communication, argued that the ELIZA and the LYRIC “chatbots” were achieving interactions that felt human to many users by exploiting human sense-making practices. In keeping with his long-term practice of using “trouble” as a way of discovering the taken-for-granted practices of human sense-making, Garfinkel designed scripts for ELIZA and LYRIC that he could disrupt in order to reveal how their success depended on human social practices. Hence, the announcement “Machine Down” by the chatbot was a desired result of Garfinkel’s interactions with it. This early (but largely unknown) research has implications not only for understanding contemporary AI chatbots, but also opens possibilities for respecifying current information systems design and computational practices to provide for the design of more flexible information objects.

A I and contemporary challenges: The good, bad and the scary

The way in which powerful AI technologies could transform our lives, society, economics, governance and most importantly ethics and morality surrounding it is quite popular in every day news media as well as our drawing room discussions at home. There is considerable confusion among individuals and communities on the impact of AI. Profit seeking global mega corporations have already poured in billions of dollars in AI research and development to maximise their profits. On the other hand, there are scary and dangerous scenarios shaping up on the impact and use of AI tools. Rouge and non-state actors are weaponizing AI technologies as well as giving rise to new fake platforms. We are not far away from killer robots, autonomous weapons and robotic war fare as depicted in the film Slaughterbots. The main purpose and objective of this essay is to understand what is good, bad and the scary of AI developments? What are the positive and negative impacts of AI on our contemporary society? Should we leave AI technology within the perspective of technological determinism or is there a scope to socially shape the new technology for the benefit of our society. These are some of the important issues that will be addressed in this essay.

Challenges to AI use in anesthesia and healthcare: An anesthesiologist’s perspective

Anesthesiology is well positioned to benefit from applications of artificial intelligence on multiple elements such as monitoring the depth of anesthesia, control of anesthetic machine functions, ultrasound guidance for procedures and diagnosis, adverse event prediction, pain assessment and management, and optimising the operating room workflow. The ethical concerns can arise from multiple aspects of AI research and deployment such as the nature and source of the data, data collection methodologies, AI models design, output interpretation and inappropriate use. AI solution can have the unintended consequences like perpetuation of systematic biases and discrimination towards under-represented sections of society. There could be conflicts about data protection, intellectual property rights and economic gains. Also, the research must be transparent and solutions feasible. The clinician’s role is ever changing in this landscape. We will discuss the broad ethical frameworks that are applicable to developing and using AI in medicine.

NEH announces major new AI research initiative

NEH announces major new AI research initiative

ACWA: an AI-driven cyber-physical testbed for intelligent water systems

Abstract This manuscript presents a novel state-of-the-art cyber-physical water testbed, namely the AI and Cyber for Water and Agriculture testbed (ACWA). ACWA is motivated by the aim to advance water resources' management using AI and cybersecurity experimentation. The main objective of ACWA is to address pressing challenges in the water and agricultural domains by utilising cutting-edge AI and data-driven technologies. These challenges include cyberbiosecurity, resources' management, access to water, sustainability, and data-driven decision-making, among others. To address such issues, ACWA is built consisting of topologies, sensors, computational clusters, pumps, tanks, smart water devices, as well as databases and AI models that control the system. Moreover, we present ACWA simulator, which is a software-based water digital twin. The simulator is based on fluid and constituent transport principles that produce a theoretical time series of a water distribution system. It creates a benchmark for comparing the theoretical approach with real-life outcomes via the physical ACWA testbed. ACWA data are available to AI and water sector researchers and are hosted in an online public repository. In this paper, the system is introduced and compared with existing water testbeds; additionally, use cases are described along with novel outcomes, such as datasets, software, and AI models.

Macine Learning Approach in Evaluating News Labels Based on Titles: Online Media Case Study

In the current digital era, information availability is abundant, and news serves as a primary source of up-to-date and reliable information for the public. However, with the increasing volume of information, a robust evaluation method is necessary to ensure accurate and dependable news labeling. This research employs a machine learning approach, utilizing three common classification algorithms: Naive Bayes, SVM, and Random Forest, to evaluate news labels based on their titles. The dataset utilized in this study is obtained from Jakarta AI Research and consists of 10,000 samples covering various news topics. Evaluation is conducted using accuracy, precision, recall, and F1-Score metrics to gain a comprehensive understanding of the classification algorithm's performance. The results of this research demonstrate that the SVM algorithm exhibits the best performance, achieving an accuracy rate of 92.92%. Random Forest follows with an accuracy rate of 91.21%, and Naive Bayes with an accuracy rate of 89.61%. These findings provide deep insights into the effectiveness of the machine learning approach in evaluating news labels based on their titles. Furthermore, the study highlights the importance of considering other evaluation metrics such as precision, recall, and F1-Score to obtain a more holistic understanding of the algorithm's performance. Further research is encouraged to involve additional classification algorithms and more diverse and extensive datasets to enhance the comprehension of news label evaluation comprehensively. Such endeavors can significantly contribute to the development of automated systems for classifying news with higher accuracy and reliability in the future

Mapping Oman's Artificial Intelligence Logistic Fuzzy Social Media Research Landscape Through Bibliometrics

Artificial intelligence and related technologies are becoming increasingly important internationally. This bibliometric study looks at Oman's research environment in AI, social media, fuzzy logic, and logistic regression from 2014 to 2024. Data for 251 English papers were obtained from Web of Science. The publication and citation patterns were examined using bibliometric approaches such as Bradford's law, Lotka's law, and network mapping. The data show a substantial increase in production and impact, particularly at younger colleges. However, national collaboration remains limited, while international participation has increased. In conclusion, Oman's emergence as an AI research hub is promising, but more concerted work and wider collaborations are required to reach its full potential. This report provides a baseline for monitoring Oman's progress toward strategic research targets. Further effort should focus on building national networks and boosting international collaboration.

Fraunhofer IGCV sets up AI-based fiber placement automated composite manufacturing

On September 27, 2023, Fraunhofer IGCV reports that the SAMBA Pro Prepreg plant of the Fraunhofer Institute for Casting, Composite and Processing Technology IGCV has found its new home in »Hall 43« of the University of Augsburg, an AI research hall inaugurated in June 2023. The production facility is now in use for research into innovative approaches in the field of machine learning. In addition, it is available to companies from various industries for active technology and application development together with researchers from the Fraunhofer Institute.

Enhancing Legal AI Research: Automatic Labeling for Legal Precedents Datasets and Evaluating Case Similarity Using BERT

Enhancing Legal AI Research: Automatic Labeling for Legal Precedents Datasets and Evaluating Case Similarity Using BERT

The energy challenges of artificial superintelligence.

We argue here that contemporary semiconductor computing technology poses a significant if not insurmountable barrier to the emergence of any artificial general intelligence system, let alone one anticipated by many to be "superintelligent". This limit on artificial superintelligence (ASI) emerges from the energy requirements of a system that would be more intelligent but orders of magnitude less efficient in energy use than human brains. An ASI would have to supersede not only a single brain but a large population given the effects of collective behavior on the advancement of societies, further multiplying the energy requirement. A hypothetical ASI would likely consume orders of magnitude more energy than what is available in highly-industrialized nations. We estimate the energy use of ASI with an equation we term the "Erasi equation", for the Energy Requirement for Artificial SuperIntelligence. Additional efficiency consequences will emerge from the current unfocussed and scattered developmental trajectory of AI research. Taken together, these arguments suggest that the emergence of an ASI is highly unlikely in the foreseeable future based on current computer architectures, primarily due to energy constraints, with biomimicry or other new technologies being possible solutions.

When cool hospitality brand meets AI: exploring the matching effect of service agents and brand images on brand attitude

PurposeThis paper aims to explore the matching effect of hospitality brand image (cool vs non-cool) and service agents (Artificial intelligence [AI] vs human staff) on brand attitude, with a focus on assessing the role of feeling right as a mediator and service failure as a moderator.Design/methodology/approachThis paper tested the hypotheses through three experiments and a Supplementary Material experiment, which collectively involved 835 participants.FindingsThe results indicated that the adoption of AI by cool brands can foster the right feeling and enhance consumers’ positive brand attitudes. In contrast, employing human staff did not lead to improved brand attitudes toward non-cool brands. Furthermore, the study found that service failure moderated the matching effect between service agents and cool brand images on brand attitude. The matching effect was observed under successful service conditions, but it disappeared when service failure occurred.Practical implicationsThe findings offer practical guidance for hospitality companies in choosing service agents based on brand image. Cool brands can swiftly transition to AI, reinforcing their modern, cutting-edge image. Traditional brands may delay AI adoption or integrate it strategically with human staff.Originality/valueTo the best of the authors’ knowledge, this paper represents one of the first studies to address the issue of selecting the optimal service agent based on hospitality brand image. More importantly, it introduces the concept of a cool hospitality brand image as a boundary condition in the framework of AI research, providing novel insights into consumers’ ambivalent responses to AI observed in previous studies.

A review of the explainability and safety of conversational agents for mental health to identify avenues for improvement.

Virtual Mental Health Assistants (VMHAs) continuously evolve to support the overloaded global healthcare system, which receives approximately 60 million primary care visits and 6 million emergency room visits annually. These systems, developed by clinical psychologists, psychiatrists, and AI researchers, are designed to aid in Cognitive Behavioral Therapy (CBT). The main focus of VMHAs is to provide relevant information to mental health professionals (MHPs) and engage in meaningful conversations to support individuals with mental health conditions. However, certain gaps prevent VMHAs from fully delivering on their promise during active communications. One of the gaps is their inability to explain their decisions to patients and MHPs, making conversations less trustworthy. Additionally, VMHAs can be vulnerable in providing unsafe responses to patient queries, further undermining their reliability. In this review, we assess the current state of VMHAs on the grounds of user-level explainability and safety, a set of desired properties for the broader adoption of VMHAs. This includes the examination of ChatGPT, a conversation agent developed on AI-driven models: GPT3.5 and GPT-4, that has been proposed for use in providing mental health services. By harnessing the collaborative and impactful contributions of AI, natural language processing, and the mental health professionals (MHPs) community, the review identifies opportunities for technological progress in VMHAs to ensure their capabilities include explainable and safe behaviors. It also emphasizes the importance of measures to guarantee that these advancements align with the promise of fostering trustworthy conversations.

Playing Various Strategies in Dominion with Deep Reinforcement Learning

Deck-building games, like Dominion, present an unsolved challenge for game AI research. The complexity arising from card interactions and the relative strength of strategies depending on the game configuration result in computer agents being limited to simple strategies. This paper describes the first application of recent advances in Geometric Deep Learning to deck-building games. We utilize a comprehensive multiset-based game representation and train the policy using a Soft Actor-Critic algorithm adapted to support variable-size sets of actions. The proposed model is the first successful learning-based agent that makes all decisions without relying on heuristics and supports a broader set of game configurations. It exceeds the performance of all previous learning-based approaches and is only outperformed by search-based approaches in certain game configurations. In addition, the paper presents modifications that induce agents to exhibit novel human-like play strategies. Finally, we show that learning strong strategies based on card combinations requires a reinforcement learning algorithm capable of discovering and executing a precise strategy while ignoring simpler suboptimal policies with higher immediate rewards.

Exploring the Function of Expressions in Negotiation: The DyNego-WOZ Corpus

For affective computing to have an impact outside the laboratory, facial expressions must be studied in rich naturalistic situations. We argue negotiations are one such situation as they are ubiquitous in daily life, often evoke strong emotions, and perceived emotion shapes decisions and outcomes. Negotiations are a growing focus in AI research and applications, including agents that negotiate directly with people and attempt to use affective information. We introduce the DyNego-WOZ Corpus, which includes dyadic negotiation between participants and wizard-controlled virtual humans. We demonstrate the value of this corpus to the affective computing community by examining participants' facial expressions in response to a virtual human negotiation partner. We show that people's facial expressions typically co-occur with the end of their partner's speech (suggesting they reflect a reaction to the content of this speech), that these reactions do not correspond to prototypical <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">emotional</i> expressions, and that these reactions can help predict the expresser's subsequent action. We highlight challenges in working with such naturalistic data, including difficulties of expression recognition during speech, and the extreme variability of expressions, both across participants and within a negotiation. Our findings reinforce arguments that facial expressions convey more than emotional state but serve important communicative functions.