The Summit supercomputer at ORNL has been the world's top AI-powered supercomputer for several years, able to handle both conventional modeling and simulation science as well as emerging AI workloads. As its mid-life approaches, this article examines the use of AI for science on Summit, reviewing its usage across a range of science projects and applications scaling AI to the full system. It also explores AI-coordinated science discovery workflows and offers observations on how AI will advance scientific knowledge and understanding, particularly in the context of leadership-class scientific computing. Overall, the article highlights Summit's impressive performance in using AI for scientific discovery, demonstrating its ability to handle complex scientific workflows and large datasets. This has enabled researchers to better model and simulate complex systems, advance materials science, and develop new drugs and therapies. The article suggests that the use of AI for science will continue to accelerate, and that leadership-class scientific computing will play a crucial role in enabling researchers to tackle ever more complex scientific challenges.

A surge in the production and usage of Internet of Things (IoT) devices has led to unprecedented security vulnerabilities in the connections between IoT, fog nodes, and cloud computing servers. In response, researchers have conducted an in-depth survey on existing intrusion detection solutions for the IoT ecosystem. The survey innovates on existing research by categorizing intrusion detection based on the approach used to detect attacks, and then further classifying each approach into a set of sub-techniques. The researchers also propose a comprehensive cybersecurity framework that combines elements of explainable artificial intelligence (XAI), federated learning, game theory, and social psychology to offer strong protections against cyberattacks. These protections are urgently needed, as IoT devices are now part of everyday life and are being used in a wide range of industries, from transportation to healthcare to smart homes. Despite their growing presence, IoT devices are resource-constrained, which means they often need to communicate with other devices to complete tasks that have large resource demands. This provides an immense opportunity for malicious parties to find vulnerabilities in the heterogeneous and multiparty architecture of IoT ecosystems, and launch attacks. The cybersecurity framework proposed by researchers aims to address these threats and protect IoT systems against cyberattacks.

The National Open University of Nigeria (NOUN) has developed a chatbot system using the Multinomial Naive Bayes algorithm to classify student complaints and inquiries. The chatbot's goal is to automate responses and ease the burden on the management system while providing students with 24/7 accessibility. In four months, the NOUN E-ticketing system generated 38,263 tickets which have been manually responded to and closed, while 7,662 tickets are still in progress. The chatbot is expected to increase student engagement, strengthen communication, and create a seamless interaction between students and their facilitators in ODL institutions, leading to a robust student-ODL relationship and ultimately resulting in a lower attrition rate. The success of the NOUN chatbot could set a precedent for other conventional higher institutions to adopt similar systems.

The author of this study analyzed 233,914 English tweets about the AI chatbot ChatGPT in its first month after launch. The tweets were analyzed using LDA topic modeling and revealed that ChatGPT can be used for news, technology, and reactions, as well as creative writing, essay writing, prompt writing, code writing, and answering questions. It was also found that ChatGPT has both positive and negative potential impacts on technologies and humans. As a result of these findings, the author discusses four key issues that need to be addressed related to this AI advancement: the evolution of jobs, a new technological landscape, the quest for artificial general intelligence, and the progress-ethics conundrum.

The OECD.AI Network of Experts has developed a Framework to Classify AI Systems that links the technical characteristics of AI systems with policy implications for the OECD Principles on Trustworthy AI. The aim of the framework is to assist policy-makers, regulators, legislators and others in assessing the opportunities and risks associated with different types of AI systems. After a year of work on the framework, a public consultation has been launched to gather feedback and input from all stakeholders including standards and technical bodies, business, legislators, regulator networks, civil society, consumers, and others. The purpose of the consultation is to assess the framework's usability and user-friendliness, as well as to gather diverse perspectives and insights. The consultation represents an opportunity for all interested parties to participate in shaping the future of AI policy and regulation.

The Domain Name System (DNS) is vulnerable to cybercriminals, making it crucial to locate and block rogue websites and their IP addresses. A solution proposed in this article is to analyze large amounts of mobile web traffic to identify dangerous domains. Text and domain traffic statistics are used to classify the domains, and three classifiers are compared for their impact. The Spark framework is used to process huge amounts of DNS traffic, which demonstrates the system's efficiency in identifying rogue domains. The MalPortrait feature was tested using real-world big ISP networks' passive DNS traffic, and the results showcased the usefulness of the feature in network security.

Small satellites such as CubeSats are being used more frequently for Earth observation missions due to their ease of development, while AI and deep neural network algorithms are enjoying success for their various applications. With this in mind, the Grenoble University Space Centre has launched the QlevEr Sat mission, which aims to embed an AI algorithm in a CubeSat to process data directly and send only relevant results to a ground station. The mission uses the Teledyne Qormino QLS1046-Space processor to monitor deforestation and create forest/non-forest and cloud/non-cloud segmentation maps from images taken with the Emerald sensor. The processor's suitability for the mission has been confirmed, and a pixel-wise classification algorithm has been designed to leverage the processor's computing power and produce highly accurate binary maps. Overall, the QlevEr Sat mission represents a promising development in the intersection of small satellite technology and AI, with potential implications for space systems going forward.

With military operations being conducted in all domains including the cyberspace, the protection of data, sensors, computer systems and other information technology infrastructure is vital. However, with a shortage in cybersecurity professionals, Artificial Intelligence Machine Learning (AI/ML) techniques are being looked to by companies, universities and government agencies to help fill the gap. The application of AI has successfully been used to detect and mitigate cyber-attacks in the Department of Defense, but adversaries have also been utilizing AI to plan and launch attacks. The Army Research Laboratory is working on employing AI services to detect and respond to attacks on a remote sensor network. Components of the framework will include AI/ML for detecting malware and intrusion detection, cloud AI tools for system monitoring, log monitoring for detecting user behavior anomalies and sensor resiliency characteristics, and a visual mapping application to display sensor and target locations. Notifications will be sent to system administrators in near real-time to keep them aware of possible threats. This paper presents a description of the components of the framework and their initial use-case for a remote sensing network application.

A new secure data dissemination scheme for healthcare systems has been proposed, using artificial intelligence (AI) and blockchain technology. In current IoT-based e-Health Systems (IoTEHS), medical devices use unsecured public communication channels to share healthcare data with nearby edge devices or cloud servers, leaving them vulnerable to exploitation by attackers. The proposed scheme would ensure the secure transfer of patient data by filtering it through an AI-based intrusion detection system located at the edge of the network, before transmitting it to centralized cloud servers where a smart contract-enabled consensus mechanism would validate the transaction, and the data would be stored on the InterPlanetary File System (IPFS) of cloud. The transaction hash would be stored on the blockchain ledger located at the edge devices for faster data exchange. The results of an experimental investigation demonstrate the proposed scheme's effectiveness and its ability to resist a variety of security attacks, making it a promising solution for securing IoTEHS.

This paper discusses the importance of constructing a knowledge graph in spacecraft launch to overcome the challenges of "sparse knowledge" and "incomplete knowledge" within a vast amount of data. By analyzing the technical architecture of a knowledge graph, the paper outlines approaches to its construction including knowledge source, modeling, extraction, fusion, inference, and storage. It is argued that a knowledge graph can be used to achieve Semantic Artificial Intelligence and improve intelligent data engineering at the launch site. The paper highlights the different applications of a knowledge graph, including automatic question-answering based on semantic search, fault detection based on machine learning and equipment or information system portraits. In conclusion, the construction and use of a knowledge graph in spacecraft launch would facilitate complex data analytics and applications, as well as being an effective method of solving sparse and incomplete knowledge problems.