A Monolithic Stochastic Computing Architecture for Energy Efficient Arithmetic.

Transient bit flips may cause unexpected behaviors in digital arithmetic circuits, and thus computation errors. It is generally believed that stochastic computing (SC) can tolerate bit flips due to its unique number representation format. In SC, a value is encoded by a random bit stream using the probability of 1s in the bit stream. Therefore, one bit flip changes the probability by a small number given a long enough bit stream, and the computation results might not change much. This is verified in previous works mostly through empirical studies. However, we found that bit flips actually change the probability of a stochastic bit stream by a predictable value. This is formulated by considering the bit flips as Bernoulli events. The distorted probability of the bit stream considering bit flips is derived and the SC bit flip models for various stochastic circuits are proposed. The error recovery based on the formulation is also proposed. The SC bit flip model and error recovery scheme are verified on basic SC elements such as the SC multipliers and an FSM-based stochastic circuit. They are then applied to a complex SC system computing a neural network. The results show that the accuracy of MNIST digit recognition can be recovered even with a high bit flip rate of 10%.

Deploying modern neural networks on resource-constrained edge devices necessitates a series of optimizations to ready them for production. These optimizations typically involve pruning, quantization, and fixed-point conversion to compress the model size and enhance energy efficiency. While these optimizations are generally adequate for most edge devices, there exists potential for further improving the energy efficiency by leveraging special-purpose hardware and unconventional computing paradigms. In this study, we explore stochastic computing neural networks and their impact on quantization and overall performance concerning weight distributions. When arithmetic operations such as addition and multiplication are executed by stochastic computing hardware, the arithmetic error may significantly increase, leading to a diminished overall accuracy. To bridge the accuracy gap between a fixed-point model and its stochastic computing implementation, we propose a novel approximate arithmetic-aware training method. We validate the efficacy of our approach by implementing the LeNet-5 convolutional neural network on an FPGA. Our experimental results reveal a negligible accuracy degradation of merely 0.01% compared with the floating-point counterpart, while achieving a substantial 27× speedup and 33× enhancement in energy efficiency compared with other FPGA implementations. Additionally, the proposed method enhances the likelihood of selecting optimal LFSR seeds for stochastic computing systems.

Stochastic Computing (SC) is a computing paradigm that allows for the\nlow-cost and low-power computation of various arithmetic operations using\nstochastic bit streams and digital logic. In contrast to conventional\nrepresentation schemes used within the binary domain, the sequence of bit\nstreams in the stochastic domain is inconsequential, and computation is usually\nnon-deterministic. In this brief, we exploit the stochasticity during switching\nof probabilistic Conductive Bridging RAM (CBRAM) devices to efficiently\ngenerate stochastic bit streams in order to perform Deep Learning (DL)\nparameter optimization, reducing the size of Multiply and Accumulate (MAC)\nunits by 5 orders of magnitude. We demonstrate that in using a 40-nm\nComplementary Metal Oxide Semiconductor (CMOS) process our scalable\narchitecture occupies 1.55mm$^2$ and consumes approximately 167$\\mu$W when\noptimizing parameters of a Convolutional Neural Network (CNN) while it is being\ntrained for a character recognition task, observing no notable reduction in\naccuracy post-training.\n

In the era of artificial intelligence, a new form of economy has appeared. Accordingly, it is necessary to adapt to the new era of technical skills, compound and innovative talents. As an important position for training technical talents, higher vocational colleges show new characteristics in the era of artificial intelligence and need to innovate the training mode of talents. Based on this, the article will analyze the difference between higher vocational education and undergraduate vocational education, and focus on the analysis of the innovative path and strategy of higher vocational education to train talents in the era of artificial intelligence, to cultivate innovative skilled talents who adapt to the era of artificial intelligence and meet the needs of the development of society and the era [1].

Artificial intelligence is an important feature and development trend of today's era, which has had a profound impact and challenge on various industries, including libraries as important institutions of knowledge service. The application of artificial intelligence technology provides new opportunities and possibilities for libraries, and also puts forward new requirements and expectations for the professional competence of librarians. Starting from the challenges and opportunities faced by librarians in the era of artificial intelligence, this article analyzes the connotation and composition of the professional competence of librarians, explores the goals, contents and approaches of professional competence construction for librarians in the era of artificial intelligence, and aims to provide references for improving the professional quality and service level of librarians and promoting the innovative development of library services.

Editorial on the Research Topic Data-intensive medicine and healthcare: ethical and social implications in the era of artificial intelligence and automated decision-making Medical "big data" and artificial intelligence (AI) are a hyped duo. Promises include developing more personalised treatments, delegating medical decision-making to tireless and seemingly objective algorithms, improving preventive screening, and providing healthcare more efficiently through predictive risk scores. AI and big data, however, do not automatically transform into improved health outcomes. The practical and functional uses of AI in big data environments require integrating and interpreting a wide variety of medical data (e.g., from genomics or other omics, imaging, biomarker analyses) and other personal data. As a result, AI-driven technology bears various new challenges and risks at the societal, algorithmic, organizational, expert, and individual levels. Scholarship on the ethical, legal, and social issues of using AI in data-intensive medicine and healthcare has highlighted numerous areas of contention, including regulation, explainability, privacy, data sharing and protection, trust, and biases, as well as how AI might affect the patient-doctor relationship and support interdisciplinary expert teams in their decisions. Aiming to extend this perspective, this Research Topic focuses on AI applications in various areas of innovative data-intensive medicine, such as genomics, neuroscience, and child and elderly care. The contributions explore how ethical and social considerations can/should be part of medical AI by considering issues of diversity, the significance of datafication and automation, public and patient participation, developing deliberative or open science approaches (such as open codes), and how to ensure interoperability among developers and users while preventing misuse, hacking, or manipulation. The Research Topic comprises 10 articles dealing with various aspects of

This literature review aims to identify the impact of innovative entrepreneurship on welfare and employment in the era of automation and artificial intelligence. Through an examination of various previous empirical and theoretical studies, rapid technological developments through automation and artificial intelligence have brought about major transformations in global entrepreneurship patterns, such as accelerating efficiency and new innovation opportunities for entrepreneurs. However, there are challenges in the form of a digital skills gap and the potential for a reduction in human-based jobs, requiring entrepreneurs to develop entrepreneurship strategies that are oriented towards human empowerment while remaining relevant in the era of automation and artificial intelligence. The results of this review confirm that entrepreneurial success in this era is not only measured by economic growth, but also by its ability to maintain a balance between technological innovation and social welfare. This study is expected to serve as the basis for the development of conceptual models and further research directions related to sustainable entrepreneurship and human welfare in the era of automation and artificial intelligence.

The era of artificial intelligence (AI) is challenging for company leaders who have to systematically observe the evolving digital environment, as well as to adopt and effectively implement new technologies within their organizations. Effective deployment of AI in a company requires full commitment of the management staff, as it plays a crucial role in shaping employee behavior. As a result, the connection between AI and leadership has increasingly been a subject of academic inquiry, although it has not yet been sufficiently examined. Therefore, the aim of this article is to determine the competencies and skills required for leaders in companies that implement AI. To fulfill this purpose, a scoping review was conducted in accordance with the PRISMA-ScR checklist and diagram flow. The study, conducted in March 2025, involved a review of articles relevant to the research topic that were indexed in the Web of Science, Springer Nature Link, and Scopus databases. The review was limited to open-access articles available in English. Out of the initial 463 articles, 28 studies were qualified for full-text review, of which 17 were selected for detailed assessment. The research led to the identification of a set of 15 competencies and skills considered essential for leaders tasked with AI implementation. The proposed recommendations are meant to improve the development of competencies and skills of leaders involved in decision-making related to AI technology implementation in companies. The findings of the study may enable more effective utilization of AI by prompting changes in leaders’ behaviors.

DeepSeek is reshaping the higher education model in all aspects by assisting differentiated teaching and personalized learning, promoting flat teaching management processes, creating interactive learning modes with deep participation, creating intelligent paths for boundaryless learning, providing accurate and comprehensive data-driven feedback, and enhancing global education equity and inclusiveness. For better applying artificial intelligence to promote the development of higher education, this study summarizes the main scenarios in which DeepSeek R1 contributes to the sustainable development of higher education. The development strategy of higher education in the era of intelligence is obtained, which includes improving teacher literacy we should continue to promote the healthy and innovative development of education models, actively deepen artificial intelligence to assist in the construction of teacher teams, enhance the interpretability of education models, improve the precision support of personalized learning, real-time detect and regulate student emotions, construct a scientific evaluation system for large models, bridge the global education digital divide, establish an ethical and moral framework for artificial intelligence, strengthen the cross-cultural adaptability of intelligent collaboration, and help lifelong learning and sustainable social development.

This comprehensive article explores the importance and role of education, knowledge and key human-centric skills in the era of Artificial Intelligence (AI). AI can be best described as an emerging technology and branch of computer science that relies on machine and deep learning in its attempt to simulate human intelligence by applying special software-encoded heuristic solutions, based on a comprehensive learning process through training data. The modern era is accompanied by numerous challenges in the application of AI, especially from the aspect of education, knowledge and key human skills that are necessary for its ethical and safe application. At its beginning, the article provides basic explanations and the history of the origin of this term, as well as the usual risks and threats of using this new technology. After that, the article is focused towards portraying the role of education and knowledge in the ubiquitous era of the application of Artificial Intelligence. In recent years, contemporary society is facing the unexpected rise of AI in its business, economic, legal, cultural, educational, economic, technological, and many other fields. It has deeply penetrated the sphere of knowledge, education and research and development (R&D) and over time it has been shown that AI has the ability to solve some of the burning problems in the education process, to introduce innovations in teaching practices, to ensure inclusive and equal quality education for all, as well as to revolutionize teaching and learning methods and scientific research processes. However, in an era of pervasive concern about whether or not AI can replace human work and existing jobs, the article concludes that while it can perform some routine tasks and procedures, it cannot replace the systems thinking, humane knowledge and key human skills, otherwise necessary for solving problems and achieving deep scientific truths, knowledge and discoveries. Finally, the article deals with an extensive analysis of key human-centric skills (creativity, critical thinking, communication and cooperation) otherwise crucial for the safe and ethical application of Artificial Intelligence systems.

The advent of the artificial intelligence (AI) era presents significant challenges to traditional education, necessitating innovative approaches to teaching methods. Integrating artificial intelligence into classroom instruction to enhance course quality has become a pressing concern. This study explores the design of macroeconomics courses in the context of artificial intelligence, addressing the limitations of traditional teaching methods and highlighting the potential benefits of AI integration. These advantages include personalized learning experiences, increased teaching efficiency, enhanced interactivity, real-time feedback, and access to expanded teaching resources. Building on these insights, the research proposes a comprehensive framework for innovating macroeconomics courses design. First, it emphasizes content innovation, which involves integrating AI-related economic phenomena and frontier developments in the discipline. Second, it focuses on methodological innovation, leveraging AI-powered tools to facilitate teaching and incorporating practical, hands-on learning experiences. Finally, it advocates for assessment reform, emphasizing process-oriented evaluations and introducing diverse, multi-dimensional assessment methods. These innovative strategies are shown to enhance the quality of macroeconomics education, improve student learning outcomes, and better align educational programs with the demands of a rapidly evolving economy in the AI era.

In the era of artificial intelligence, traditional physical education methods can no longer meet the needs of contemporary college physical education, and it is even more difficult to adapt to the development needs of physical education in the new era. And through the new era of human artificial intelligence technology, physical education can be gradually reformed, and artificial intelligence technology can be reasonably used to construct an intelligent robot system for physical education. The content, scenes, methods and processes of physical education can be continuously innovated and diversified. The evaluation system can better adapt to the individual development needs of contemporary college students. Artificial intelligence makes physical education more intelligent, curriculum construction more scientific, and teaching methods more innovative and future.

This article focuses on analyzing fathers' communication patterns in shaping early childhood science skills in the era of Artificial Intelligence (AI). The main problem identified is the changing dynamics of communication between fathers and children, and its potential impact on the development of children's science skills. The purpose of this study is to identify effective father communication patterns in stimulating children's interest and science skills in the AI era. The research method was qualitative case study of five fathers. Data were collected through interviews, observations, and content analysis of father-child interactions in the context of science learning. Then analyzed using the Miles Huberman model and then data validity with data triangulation. The results showed that father's communication focused on open-ended questions, presenting information with a creative approach, and using AI technology as a learning tool can improve children's science skills. Fathers' understanding and support of technological development is also a key factor in building children's science literacy in the AI era. The findings contribute to practical understanding for parents, educators and policy makers to shape a communication environment that supports the development of children's science skills. The practical implications can be applied in designing family education programs and science learning strategies for early childhood that are appropriate to the development of AI technology.

The artificial intelligence (AI) era has posed a great challenge to the cultivation of labour values in schools and universities, while labour values, whose theoretical logic originates from the dialectical unity relationship between the subject and object of labour, are the core in labour education. Labour values are embodied by labour subjectivity and developed by labour practicality. The research takes the bidirectional relationship between AI and labour education as the theoretical perspective to locate three major problems confronted by labour education: (1) educational administration in schools and universities; (2) separation between technique and morality; (3) binary opposition between mental and manual labour. In this case, labour education in schools and universities should adhere to the proper value orientation of labour education, promote the inter-coordination between labour skills and labour values, and pursue a system that integrates the five education dimensions—morality, intelligence, physique, aesthetics, and labour.

In the era of artificial intelligence (AI), physical education (PE) teachers face unprecedented challenges in integrating advanced technologies into their teaching practices. This study aims to identify the specific challenges PE teachers encounter in the AI era and propose practical strategies to enhance the quality and relevance of school-based physical education. This study employed a multi-method approach, including literature review, mathematical statistics, questionnaire surveys, and expert interviews. The participants were 120 PE teachers from urban and rural schools across different regions. The study began with a literature review to establish a theoretical foundation. Questionnaires were used to gather teachers' perceptions of challenges related to AI integration, student engagement, and professional development. In-depth interviews with 25 PE experts provided deeper insights into their experiences and coping strategies. Data were analyzed using SPSS software. The results showed that 80% of PE teachers reported difficulties in integrating AI tools into their lessons, while 70% struggled with engaging students in technology-enhanced activities. Analysis of the questionnaire and interview data revealed three main challenges: (1) limited access to AI resources, (2) lack of training in AI applications, and (3) concerns about the impact of AI on traditional teaching methods. Teachers emphasized the need for continuous professional development and support to effectively utilize AI in teaching. The study concludes that PE teachers face significant challenges in adapting to the AI-driven educational environment, particularly in integrating AI tools and maintaining student engagement. These findings align with previous research on educational technology integration but highlight the unique concerns of PE teachers regarding the impact of AI on traditional teaching. This study proposes the following strategies: (1) increasing investment in AI tools and establishing resource-sharing platforms to improve access to AI resources; (2) providing continuous professional training in AI applications and integrating AI education into teacher development programs; (3) offering technical support for AI integration and encouraging teachers to use AI tools to optimize traditional teaching methods.