AI Domain Research Articles

TS-MAE: A masked autoencoder for time series representation learning

Self-supervised learning (SSL) has been widely researched in recent years. In Particular, generative self-supervised learning methods have achieved remarkable success in many AI domains, such as MAE in computer vision, well-known BERT, GPT in natural language processing, and GraphMAE in graph learning. However, in the context of time series analysis, not only is the work that follows this line limited but also the performance has not reached the potential as promised in other fields. To fill this gap, we propose a simple and elegant masked autoencoder for time series representation learning. Firstly, unlike most existing work which uses the Transformer as the backbone, we build our model based on neural ordinary differential equation which possesses excellent mathematical properties. Compared with the position encoding in Transformer, modeling the evolution patterns continuously could better extract the temporal dependency. Secondly, a timestamp-wise mask strategy is provided to cooperate with the autoencoder to avoid bias, and it also could reduce the cross-imputation between variables to learn more robust representations. Lastly, extensive experiments conducted on two classical tasks demonstrate the superiority of our model over the state-of-the-art ones.

Is There a Domain of Linguistic Competence AI Cannot Grasp?

Abstract Linguistic competence is, among other things, the cognitive ability of using words appropriately. Following this notion, being a competent language user cannot be merely simulated because pretending to use words appropriately already is using words appropriately. Since current AI already can use words appropriately it may be that AI already has linguistic competence instead of merely simulating it. However, in this article I argue that there is at least one domain of linguistic competence that can be simulated: grasping lexical effects. Since grasping lexical effects presupposes an ability which AI lacks, namely feeling emotions, I argue that AI can only grasp the cognitive part of words but not their lexical effects. AI can only simulate to grasp lexical effects. Hence, lexical effects are a domain of linguistic competence AI cannot grasp.

Toward the Dynamic Relationship Between AI Transparency and Trust in AI: A Case Study on ChatGPT

This study examines the relationship between transparency and trust in AI. Traditional interpersonal trust attributes are considered mediators between transparency and trust. However, in the context of AI, the role of trust attributes has been redefined. Transparency is divided into accuracy, clarity, and disclosure, whereas trust in AI is identified through functionality, helpfulness, and predictability. The survey analysis and experimental evidence revealed the dynamic relationships between these factors. Accuracy and clarity influence the overall trust in AI by alleviating distrust only when interacting with functionality and helpfulness, respectively. Furthermore, predictability serves as a mediator, diminishing the mistrust between AI transparency and the overall trust in AI. The human capability to interpret information provided by AI transparency enhances this mediation. This study not only enhances understanding within the AI domain but also contributes significantly to the nuanced comprehension of the complex relationship between AI transparency and trust in AI, expanding previous simplistic interpretations.

Design Insights for Developing Deepfakes Awareness Games for Adolescents

Around a decade ago, significant advancements in the deep learning domain like Generative Adversarial Networks (GANs) started to facilitate the creation of new synthetical content. The development of deepfake algorithms gained momentum due to the increase in computing power and open-source available resources which facilitated broader access to advanced deep learning methods. While initially deepfakes were created by creative pursuits in various media and marketing campaigns, they started to be misused driven by malicious purposes in various social media manipulation campaigns. These campaigns imply spreading disinformation and misinformation for discrediting individuals and creating confusion in relation to specific topics and events and are meant to altering users’ beliefs and behaviours. Such facts raised important ethical and societal concerns. Although deepfakes still represent a recent research area, a large body of studies is dedicated to, on the one hand, generation, and detection of deepfakes, and on the other hand, to understanding their implications and consequences for society. At the same time, governmental and practitioner efforts are devoted to containing their use and impact on society by proposing various strategies and programs. Nevertheless, both academic and societal efforts are in an incipient stage and often adopt a generalist perspective. This represents a crucial point in relation to the fact that many users in various social media platforms like TikTok is represented by adolescents that are known to be unaware and more vulnerable to media content in general, and deepfakes content, in particular. This represents the knowledge gap that this research aims to tackle by synthesizing design insights as requirements and guidelines that can be used when developing and deploying deepfakes awareness games for producing and/or enhancing awareness of adolescents. Accordingly, a systematic literature review is conducted and merged with previous experience in this domain by adopting a transdisciplinary research approach that merges methods and techniques from the deepfakes, AI, social media, cyber security, and gamification domains.

Towards AI driven surface roughness evaluation in manufacturing: a prospective study

AbstractIn the era of Industry 4.0 and the digital transformation of the manufacturing sector, this article explores the significant potential of machine learning (ML) and deep learning (DL) techniques in evaluating surface roughness—a critical metric of product quality. The integration of edge computing with current computational resources and intelligent sensors has revolutionized the application of AI-driven algorithms in smart manufacturing. It provides real-time data analysis and decision-making capabilities that were unattainable only a decade ago. The research effort intends to improve data-driven decision-making for product quality evaluation by leveraging data integration from manufacturing operations and surface quality measurements. Although a substantial amount of research has been conducted in the related fields, it is still difficult to comprehend and compile all the data on surface roughness research predictive assessment in the form of a process pipeline. This thorough systematic analysis examines scholarly articles published between 2014 and 2024 focusing on surface roughness assessment in precision manufacturing settings. The article is thoroughly classified based on the manufacturing processes, datasets, and ML models used, giving light on the present status, prominent approaches, and existing issues in this sector. A table summarizing the relevant works in this domain providing an easy access to the current trends have been provided. The article not only compiles essential findings and identifies research gaps and similarities in existing methodologies, but it also discusses future research directions and open issues in AI-aided surface roughness evaluation. The critical analysis of the literature reveals a scientific gaps which includes consistent development of benchmarked datasets and making the AI models more explainable using the state-of-the-art explainable AI (XAI) algorithms. The ultimate objective of the article is not only to provide a guide for the practitioners in either of the three domains of AI, manufacturing or surface metrology but also to pave the path for more robust, efficient, and accurate surface quality evaluation processes in production.

Beyond the code: The impact of AI algorithm transparency signaling on user trust and relational satisfaction

This study investigates the effectiveness of AI-algorithm transparency signaling as a strategy to enhance organization-public relationships (OPRs) in AI-assisted communications. Building upon signaling theory and trust transfer theory, the study examines whether the AI algorithm transparency influences trust in AI systems by users and this trust can be transferred into trust in AI systems’ parent company, which in turn, influences the relational satisfaction with the company. An online experiment with 537 participants demonstrated that transparency signaling significantly improves users’ relational satisfaction with the AI parent company. However, this effect is mediated by trust in both the AI system and the parent company, rather than a direct relationship. Our findings offer practical guidelines for AI domain experts and public relations practitioners to deliberately convey the true essence of transparency in AI-mediated communication and ensure accountability in AI adoption, thereby improving public relations outcomes.

Analyzing the Effects of Transcription Errors on Summary Generation of Bengali Spoken Documents

Automatic speech recognition (ASR) has become an indispensable part of the AI domain, with various speech technologies reliant on it. The quality of speech recognition depends on the amount of annotated data used to train an ASR system, among other factors. For a low-resourced language, this is a severe constraint and thus ASR quality is often poor. Humans can read through text containing ASR-errors, provided the context of the sentence is preserved. Yet in cases of transcripts generated by ASR systems of low-resource languages, multiple important words are misrecognized and the context is mostly lost; discerning such a text becomes nearly impossible. This paper analyzes the types of transcription errors that occur while generating ASR transcripts of spoken documents in Bengali, an under-resourced language predominantly spoken in India and Bangladesh. The transcripts of the Bengali spoken document are generated using the ASR of Google Cloud Speech. The paper also explores if there is an effect of such transcription errors in generating speech summaries of these spoken documents. Summarization is carried out extractively; sentences are selected from the ASR-generated text of the spoken document. Speech summaries are created by aggregating the speech-segments from the original speech of the selected sentences. Subjective evaluation shows the ‘readability’ of the spoken summaries are not degraded by ASR errors, but the quality is affected due to the reliance on intermediate text-summary containing transcription errors.

Exploring the predictors of public acceptance of artificial intelligence-based resurrection technologies

Given their novelty and ethical complexity, this study delves into the public acceptance of artificial intelligence-based resurrection technologies (AI-RTs), an emerging area in the AI domain that proposes to digitally “resurrect” individuals who have passed away. Employing a survey-based experimental design, the study explores various cognitive, affective, normative, and ethical predictors of acceptance, as outlined in the Technology Acceptance Model and its extensions. A nationally representative sample of U.S. adults (N = 1115) was randomly exposed to the description of an AI-RT application — virtual reality, chatbot, or deepfake — to gauge variations in public attitude and behavioral intention. The findings reveal a nuanced understanding of public sentiment towards AI-RTs. Factors such as perceived usefulness, perceived ease of use, perceived benefit, positive emotions, and negative emotions emerged as significant influencers of both attitude towards and intention to use AI-RTs. This study contributes to the understanding of public acceptance of controversial and ethically charged technologies, offering insights for developers, marketers, media, and regulators.

Research on data preprocessing method for artificial intelligence algorithm based on user online behavior

This paper delves into the significance and efficacy of data preprocessing techniques specifically tailored for user online behavior data in enhancing the performance of artificial intelligence algorithms. Through a comprehensive literature review, we identify gaps and opportunities in current methodologies, setting the stage for the development of a novel data preprocessing approach. This proposed method is meticulously designed to handle the complexities and nuances of user online behavior data. We conduct a series of case studies and rigorous empirical analyses to evaluate the effectiveness of our approach. The results clearly demonstrate that our method substantially improves data quality by effectively reducing noise and eliminating irrelevant information, which, in turn, enhances the overall performance of the AI algorithms. The paper concludes with a discussion on the limitations of the current study and provides insightful directions for future research in the field. This includes potential refinements to the preprocessing technique and its application to other types of behavioral data in different AI domains.

How generative AI will drive enterprise innovation

PurposeMost recent C-suite surveying suggests current applications of generative AI, although hyped, are fragmented and unlikely to yield major financial returns anticipated. Instead, business leaders expect major value from generative AI will be achieved through application of generative AI to innovation: operational innovation, product and service innovation, and most elusive of all, business model innovation.Design/methodology/approachFindings and analysis presented draws on data from several surveys of C-level executives conducted by IBM Institute for Business Value in collaboration with Oxford Economics during 2023. Each survey focused on the potential of generative AI in a particular business area. The n-count of each survey ranged from 100-3000.Findings1. Business leaders expect generative AI to build on returns achieved from investments in traditional AI, with 10 percent RoI expected on generative AI investments by 2025. 2. Executives anticipate that generative AI will have most impact when implemented to expand innovation. 3. Specific examples provided for operational innovation, product innovation, and business model innovationResearch limitations/implicationsWe are still very early in the generative AI development cycle. We have made best efforts to project, but only time will tell for sure.Practical implicationsBusiness application of generative AI are extremely fragmented. Despite the desire to throw investments at the wall to see what sticks, it is important that leaders take a structured approach to generative AI, focusing on RoI from innovation investments.Social implicationsTo alleviate negative impacts of generative AI, focusing on innovation potential and value maximization is crucial.Originality/valueThis research is based on completely new surveying and data. This papers adds to the sum total of new knowledge in the generative AI domain.

HOMMIE: CONVERSATIONAL AI ASSISTANT

Conversational AI assistants have witnessed remarkable progress in recent years, transforming human-computer interaction across a spectrum of applications. This paper offers a comprehensive overview of the state-of-the-art techniques, methodologies, and challenges in the field. We examine the core components of conversational AI, including natural language understanding, dialogue management, and response generation. Additionally, we address key challenges such as context modeling, personalization, and ethical considerations. The paper serves as a roadmap for researchers and developers, highlighting current achievements and avenues for future advancements in the dynamic domain of conversational AI. keyword - Conversational AI, NLP, Assistant, Machine Learning, Chatbot

Evaluation of large language models in breast cancer clinical scenarios: a comparative analysis based on ChatGPT-3.5, ChatGPT-4.0, and Claude2.

Large language models (LLMs) have garnered significant attention in the AI domain owing to their exemplary context recognition and response capabilities. However, the potential of LLMs in specific clinical scenarios, particularly in breast cancer diagnosis, treatment, and care, has not been fully explored. This study aimed to compare the performances of three major LLMs in the clinical context of breast cancer. In this study, clinical scenarios designed specifically for breast cancer were segmented into five pivotal domains (nine cases): assessment and diagnosis, treatment decision-making, postoperative care, psychosocial support, and prognosis and rehabilitation. The LLMs were used to generate feedback for various queries related to these domains. For each scenario, a panel of five breast cancer specialists, each with over a decade of experience, evaluated the feedback from LLMs. They assessed feedback concerning LLMs in terms of their quality, relevance, and applicability. There was a moderate level of agreement among the raters (Fleiss' kappa=0.345, P<0.05). Comparing the performance of different models regarding response length, GPT-4.0 and GPT-3.5 provided relatively longer feedback than Claude2. Furthermore, across the nine case analyses, GPT-4.0 significantly outperformed the other two models in average quality, relevance, and applicability. Within the five clinical areas, GPT-4.0 markedly surpassed GPT-3.5 in the quality of the other four areas and scored higher than Claude2 in tasks related to psychosocial support and treatment decision-making. This study revealed that in the realm of clinical applications for breast cancer, GPT-4.0 showcases not only superiority in terms of quality and relevance but also demonstrates exceptional capability in applicability, especially when compared to GPT-3.5. Relative to Claude2, GPT-4.0 holds advantages in specific domains. With the expanding use of LLMs in the clinical field, ongoing optimization and rigorous accuracy assessments are paramount.

Natural Language Processing Patents Landscape Analysis

Understanding NLP patents provides valuable insights into innovation trends and competitive dynamics in artificial intelligence. This study uses the Lens patent database to investigate the landscape of NLP patents. The overall patent output in the NLP field on a global scale has exhibited a rapid growth over the past decade, indicating rising research and commercial interests in applying NLP techniques. By analyzing patent assignees, technology categories, and geographic distribution, we identify leading innovators as well as research hotspots in applying NLP. The patent landscape reflects intensifying competition between technology giants and research institutions. This research aims to synthesize key patterns and developments in NLP innovation revealed through patent data analysis, highlighting implications for firms and policymakers. A detailed understanding of NLP patenting activity can inform intellectual property strategy and technology investment decisions in this burgeoning AI domain.

Human-Centered AI in Military Cyber Operations

Military Cyber Operations are an integral part of modern warfare and national security strategies as they crossed the science fiction realm, represent a real operational battlefield, and developed into an option in military toolboxes. Seeing ongoing technological advancements that allow the creation and use of complex mechanisms and technologies, the increasing digitalization of critical infrastructure, the growing abundance of data collected, generated, and exchanged between multiple parties, and the rise of stakeholders engaging in building and/or executing military Cyber Operations together with the increased number of such operations being conducted all over the globe reflects important lessons that need to be learned: the core element of such operations are humans: they build, acquire, execute, and assess them while also being impacted entities by them, e.g., through (psychological or physical) injury or death, or damage or destructions of human infrastructure. Moreover, the key process governing all the life cycle phases of military Cyber Operations is human decision-making or humanly assisted/augmented decision-making relying on advanced intelligent methods built with AI. Nevertheless, building and conducting military Cyber Operations should be done in a legal, responsible, and effective way implying a deep understanding of the context and adversary, proper target and cyber weapon selection, development, and use, and a clear overview of potential effects produced. These represent important aspects that should be properly defined and tackled in this domain. Hence, this research aims to introduce the Human-Centred AI concept and approach in the military cyber domain to illustrate ways to prioritize human involvement and interaction, human understanding, effective decision-making, and ethical considerations when building and conducting military Cyber Operations. To this end, an extensive literature review is conducted in the military, cyber, and AI domains together with instantiation on military Cyber Operations.

Generative AI: An In-depth Exploration of Methods, Uses, and Challenges

Recently, artificial intelligence has surged to the forefront of computer science, with generative AI emerging as the most sought-after research area. The success of generative AI hinges on advancements in algorithms, training frameworks, and data. Among these, training algorithms are of paramount importance. This article aims to shed light on several leading training algorithms in the domain. Generative AI has left a profound imprint on a myriad of industries. Intelligent publishing, advertising content creation, and finance are just a few sectors that have been revolutionized by this technology. As with all significant technological shifts, generative AI is not without its challenges. The implications it holds for employment, intellectual property rights, as well as security and privacy concerns, are profound. It's vital for stakeholders in the AI domain and beyond to consider and address these challenges. As generative AI continues to integrate more deeply into industries and our daily lives, proactive steps need to be taken to ensure ethical, secure, and equitable use. This not only guarantees the continued growth and trust in the technology but also safeguards society's values and norms in the face of rapid innovation.

Evolution and Efficiency in Neural Architecture Search: Bridging the Gap between Expert Design and Automated Optimization

The paper provides a comprehensive overview of Neural Architecture Search (NAS), emphasizing its evolution from manual design to automated, computationally driven approaches. It covers the inception and growth of NAS, highlighting its application across various domains, including medical imaging and natural language processing. The document details the shift from expert-driven design to algorithm-driven processes, exploring initial methodologies like reinforcement learning and evolutionary algorithms. It also discusses the challenges of computational demands and the emergence of efficient NAS methodologies, such as Differentiable Architecture Search and hardware-aware NAS. The paper further elaborates on NAS's application in computer vision, NLP, and beyond, demonstrating its versatility and potential for optimizing neural network architectures across different tasks. Future directions and challenges, including computational efficiency and the integration with emerging AI domains, are addressed, showcasing NAS's dynamic nature and its continued evolution towards more sophisticated and efficient architecture search methods.

ChatGPT or Bard: Who is a better Certified Ethical Hacker?

In this study, we compare two leading Generative AI (GAI) tools, ChatGPT and Bard, specifically in Cybersecurity, using a robust set of standardized questions from a validated Certified Ethical Hacking (CEH) dataset. In the rapidly evolving domain of Generative AI (GAI) and large language models (LLM), a comparative analysis of tools becomes essential to measure their performance. We determine the Comprehensiveness, Clarity, and Conciseness of the AI-generated responses through a detailed questioning-based framework. The study revealed an overall accuracy rate of 80.8 % for ChatGPT and 82.6 % for Bard, indicating comparable capabilities and specific differences. Bard slightly outperformed ChatGPT in accuracy, while ChatGPT exhibited superiority in Comprehensiveness, Clarity, and Conciseness of responses. Introducing a confirmation query like “Are you sure?” increased accuracy for both generative AI tools, illustrating the potential of iterative query processing in enhancing GAI tools' effectiveness. The readability evaluation placed both tools at a college reading level, with Bard marginally more accessible. While evaluating certain questions, a distinct pattern emerged where Bard provided generic denials of assistance while ChatGPT referenced “ethics.” This discrepancy illustrates the contrasting philosophies of the developers of these tools, with Bard possibly following stricter guidelines, especially in sensitive topics like Cybersecurity. We explore the implications and identify key areas for future research that become increasingly relevant as GAI tools see broader adoption.

Dual use concerns of generative AI and large language models

ABSTRACT We suggest the implementation of the Dual Use Research of Concern (DURC) framework, originally designed for life sciences, to the domain of generative AI, with a specific focus on Large Language Models (LLMs). With its demonstrated advantages and drawbacks in biological research, we believe the DURC criteria can be effectively redefined for LLMs, potentially contributing to improved AI governance. Acknowledging the balance that must be struck when employing the DURC framework, we highlight its crucial political role in enhancing societal awareness of the impact of generative AI. As a final point, we offer a series of specific recommendations for applying the DURC approach to LLM research.

Research on performance evaluation of higher vocational education informatization based on data envelopment analysis

&lt;abstract&gt; &lt;p&gt;This article highlights the multifaceted role of AI in modern education and offers insights into innovative ways to revolutionize educational practices through AI technologies. Since this article provides comprehension of the scope and depth of AI's impact on the education sphere, it appeals to a diverse readership, encompassing educators, policymakers, researchers, and the general public. This article explores key issues within the domain of AI in education, including personalized learning, AI-driven assessments, data analytics, and the integration of AI into learning management systems. The article highlights promises, potentials, and challenges accompanying this technological advancement. The authors emphasize the need for a balanced and informed approach to using AI to enhance the education system.&lt;/p&gt; &lt;/abstract&gt;

The Knowledge Structure and Development Trend in Artificial Intelligence Based on Latent Feature Topic Model

Currently, with the rapid development of science and technology, the field of artificial intelligence presents characteristics such as a wide crossover of disciplines and fast update, and the field of artificial intelligence has become a new focus of international competition. As an interdisciplinary field, the field of artificial intelligence has rich knowledge and strategic management significance. This article conducts an in-depth study on the knowledge structure and evolution trends in the field of AI, and the main work is as follows. First, a new potential feature topic model New-LDA is proposed for the study of topic recognition, which enhances the feature learning ability of the traditional LDA model, and makes up for the deficiency of the traditional LDA model in the ability of recognizing topics in complex environments. Second, the knowledge structure in the field of AI is analyzed from two aspects: topic recognition and coword analysis. The time series model is introduced to establish the topic evolution network, and the high-frequency words in three periods are compared and analyzed to find the evolution regular of knowledge structure in the AI domain. Finally, taking the cross-discipline of AI as an example, the thematic evolution of the field and its cross-discipline is analyzed to determine the future development direction and evolutionary trend of the field of AI.