AI Engine Research Articles

A Comprehensive Review of AI Advancement Using testFAILS and testFAILS-2 for the Pursuit of AGI

In a previous paper we defined testFAILS, a set of benchmarks for measuring the efficacy of Large Language Models in various domains. This paper defines a second-generation framework, testFAILS-2 to measure how current AI engines are progressing towards Artificial General Intelligence (AGI). The testFAILS-2 framework offers enhanced evaluation metrics that address the latest developments in Artificial Intelligence Linguistic Systems (AILS). A key feature of this re-view is the “Chat with Alan” project, a Retrieval-Augmented Generation (RAG)-based AI bot inspired by Alan Turing, designed to distinguish between human and AI generated interactions, thereby emulating Turing’s original vision. We assess a variety of models, including ChatGPT-4o-mini and other Small Language Models (SLMs), as well as prominent Large Language Models (LLMs), utilizing expanded criteria that encompass result relevance, accessibility, cost, multimodality, agent creation capabilities, emotional AI attributes, AI search capacity, and LLM-robot integration. The analysis reveals that testFAILS-2 significantly enhances the evaluation of model robustness and user productivity, while also identifying critical areas for improvement in multimodal processing and emotional reasoning. By integrating rigorous evaluation standards and novel testing methodologies, testFAILS-2 advances the assessment of AILS, providing essential insights that contribute to the ongoing development of more effective and resilient AI systems towards achieving AGI.

RAI Guidelines: Method for Generating Responsible AI Guidelines Grounded in Regulations and Usable by (Non-)Technical Roles

Many guidelines for responsible AI have been suggested to help AI practitioners in the development of ethical and responsible AI systems. However, these guidelines are often neither grounded in regulation nor usable by different roles, from developers to decision makers. To bridge this gap, we developed a four-step method to generate a list of responsible AI guidelines; these steps are: (1) manual coding of 17 papers on responsible AI; (2) compiling an initial catalog of responsible AI guidelines; (3) refining the catalog through interviews and expert panels; and (4) finalizing the catalog. To evaluate the resulting 22 guidelines, we incorporated them into an interactive tool and assessed them in a user study with 14 AI researchers, engineers, designers, and managers from a large technology company. Through interviews with these practitioners, we found that the guidelines were grounded in current regulations and usable across roles, encouraging self-reflection on ethical considerations at early stages of development. This significantly contributes to the concept of 'Responsible AI by Design'- a design-first approach that embeds responsible AI values throughout the development lifecycle and across various business roles.

Comparative Analysis of Next-Generation Space Computing Applications on AMD-Xilinx Versal Architecture

Space computing has unique considerations that limit the achievable performance capabilities of onboard processing. Due to these limiting factors, current state-of-the-art devices for space computing are unable to meet the performance and energy-efficiency requirements for next-generation communication, navigation, and artificial-intelligence applications planned for future science and exploration missions. Space designers are transitioning to domain-specific architectures with specialized acceleration hardware, such as the AMD-Xilinx Versal Adaptive System-on-Chip, to address these issues. This heterogeneous platform provides developers with several processing subsystems that allow for mission-specific customization, including scalar processing units, programmable logic, and AI engine technology enabled by vector processing units. In this paper, performance, power-consumption, energy-efficiency, and resource-utilization tradeoffs for each Versal processing subsystem are evaluated by benchmarking a suite of representative next-generation artificial-intelligence and communication applications for space. This evaluation yields a design tradespace that mission designers can use to determine the Pareto-optimal solution for an artificial-intelligence or communication application based on performance requirements and power constraints. Additionally, this evaluation demonstrates that the performance and energy-efficiency benefits of the AI engines for space computing are significant; however, they are curtailed by considerably more power consumption than the scalar processors and programmable logic.

Exploring the synergy between generative AI and software engineering: Automating code optimization and bug fixing

As applied to software engineering, generative AI is quickly transitioning from a zero-sum industry game changer into the primary automation tool for code optimization, bug identification, and problem-solving. This technology takes advantage of artificial intelligence algorithms within machine learning models to analyze and write code, resulting in improved quality and speed of an application development process. The generative AI replenishes productivity in development work and enhances centralization between development work teams through code handling and intelligent suggestions for essential codes. However, the integration of AI in software engineering poses the following problems and ethical questions: the question of accuracy, bias, and data. This paper will review the existing knowledge on generative AI in software engineering regarding its current use, future evolutions and advancements, issues and limitations, and ethical factors in using this technology. This paper considers these aspects to give a global outlook on how generative AI will transform software development in the future and how responsible AI should be employed.

SEA4DQ 2024 Workshop Summary

Welcome to the sixth edition of the workshop on Machine Learning Techniques for Software Quality Evaluation (SEA4DQ 2024), held in Brazil, July 16th, 2024, co-located with ESEC / FSE 2024 [1]. Three papers from all over the world were submitted, and all of them were accepted based on the overall positive reviews. The program also featured two keynotes by Denys Poshyvanyk on Towards an Interpretable Science of Deep Learning for Software Engineering: A Causal Inference View and Qinghua Lu on Responsible AI Engineering from A Data Perspective.

CHARM 2.0: Composing Heterogeneous Accelerators for Deep Learning on Versal ACAP Architecture

Dense matrix multiply (MM) serves as one of the most heavily used kernels in deep learning applications. To cope with the high computation demands of these applications, heterogeneous architectures featuring both FPGA and dedicated ASIC accelerators have emerged as promising platforms. For example, the AMD/Xilinx Versal ACAP architecture combines general-purpose CPU cores and programmable logic with AI Engine processors optimized for AI/ML. An array of 400 AI Engine processors executing at 1 GHz can provide up to 6.4 TFLOPS performance for 32-bit floating-point (FP32) data. However, machine learning models often contain both large and small MM operations. While large MM operations can be parallelized efficiently across many cores, small MM operations typically cannot. We observe that executing some small MM layers from the BERT natural language processing model on a large, monolithic MM accelerator in Versal ACAP achieved less than 5% of the theoretical peak performance. Therefore, one key question arises: How can we design accelerators to fully use the abundant computation resources under limited communication bandwidth for end-to-end applications with multiple MM layers of diverse sizes? We identify the biggest system throughput bottleneck resulting from the mismatch between the massive computation resources of one monolithic accelerator and the various MM layers of small sizes in the application. To resolve this problem, we propose the CHARM framework to compose multiple diverse MM accelerator architectures working concurrently on different layers within one application. CHARM includes analytical models that guide design space exploration to determine accelerator partitions and layer scheduling. To facilitate system designs, CHARM automatically generates code, enabling thorough onboard design verification. We deploy the CHARM framework on four different deep learning applications in FP32, INT16, and INT8 data types, including BERT, ViT, NCF, and MLP, on the AMD/Xilinx Versal ACAP VCK190 evaluation board. Our experiments show that we achieve 1.46 TFLOPS, 1.61 TFLOPS, 1.74 TFLOPS, and 2.94 TFLOPS inference throughput for BERT, ViT, NCF, and MLP in FP32 data type, respectively, which obtain 5.29 \(\times\) , 32.51 \(\times\) , 1.00 \(\times\) , and 1.00 \(\times\) throughput gains compared to one monolithic accelerator. CHARM achieves the maximum throughput of 1.91 TOPS, 1.18 TOPS, 4.06 TOPS, and 5.81 TOPS in the INT16 data type for the four applications. The maximum throughput achieved by CHARM in the INT8 data type is 3.65 TOPS, 1.28 TOPS, 10.19 TOPS, and 21.58 TOPS, respectively. We have open-sourced our tools, including detailed step-by-step guides to reproduce all the results presented in this article and to enable other users to learn and leverage CHARM framework and tools in their end-to-end systems: https://github.com/arc-research-lab/CHARM .

Socially Disruptive Technologies and Conceptual Engineering

In this special issue, we focus on the connection between conceptual engineering and the philosophy of technology. Conceptual engineering is the enterprise of introducing, eliminating, or revising words and concepts. The philosophy of technology examines the nature and significance of technology. We investigate how technologies such as AI and genetic engineering (so-called “socially disruptive technologies”) disrupt our practices and concepts, and how conceptual engineering can address these disruptions. We also consider how conceptual engineering can enhance the practice of ethical design. The issue features seven articles that discuss a range of topics, including trust in blockchain applications and the evolving concept of nature. These articles highlight that as technology changes the world and our concepts, conceptual engineering provides invaluable tools and frameworks to reflect on these changes and adapt accordingly.

It Takes a Village

The introduction of Large Language Models (LLM) to the chatbot landscape has opened intriguing possibilities for academic libraries to offer more responsive and institutionally contextualized support to users, especially outside of regular service hours. While a few academic libraries currently employ AI-based chatbots on their websites, this service has not yet become the norm and there are no best practices in place for how academic libraries should launch, train, and assess the usefulness of a chatbot. In summer 2023, staff from the University of Delaware’s Morris Library information technology (IT) and reference departments came together in a unique partnership to pilot a low-cost AI-powered chatbot called UDStax. The goals of the pilot were to learn more about the campus community’s interest in engaging with this tool and to better understand the labor required on the staff side to maintain the bot. After researching six different options, the team selected Chatbase, a subscription-model product based on ChatGPT 3.5 that provides user-friendly training methods for an AI model using website URLs and uploaded source material. Chatbase removed the need to utilize the OpenAI API directly to code processes for submitting information to the AI engine to train the model, cutting down the amount of work for library information technology and making it possible to leverage the expertise of reference librarians and other public-facing staff, including student workers, to distribute the work of developing, refining, and reviewing training materials. This article will discuss the development of prompts, leveraging of existing data sources for training materials, and workflows involved in the pilot. It will argue that, when implementing AI-based tools in the academic library, involving staff from across the organization is essential to ensure buy-in and success. Although chatbots are designed to hide the effort of the people behind them, that labor is substantial and needs to be recognized.

The importance of image quality in AI visual inspection systems

AbstractThe use of AI for visual inspection is currently making great strides in sectors like automotive, metals and machinery manufacturing. However, those investing in the technology not only expect a system that will be superior to manual inspection, but an error rate close to zero in defect detection. Let us look at the importance of image quality and why the variability in the production environment is the single greatest challenge that AI engineers must face.

The shutdown problem: an AI engineering puzzle for decision theorists

AbstractI explain and motivate the shutdown problem: the problem of designing artificial agents that (1) shut down when a shutdown button is pressed, (2) don’t try to prevent or cause the pressing of the shutdown button, and (3) otherwise pursue goals competently. I prove three theorems that make the difficulty precise. These theorems suggest that agents satisfying some innocuous-seeming conditions will often try to prevent or cause the pressing of the shutdown button, even in cases where it’s costly to do so. I end by noting that these theorems can guide our search for solutions to the problem.

An ontology for automated fault detection & diagnostics of HVAC using BIM and machine learning concepts

This paper presents an ontology for AFDD (Automated Fault Detection and Diagnostics) of HVAC (Heating, Ventilation, and Air conditioning) systems in buildings called “AFDDOnto”. Presently, the AFDD models are mainly data-centric and often lack semantic information such as contextual information and spatial information; additionally, configuration information, analysis, and results used for model development are lost once developed. This impedes an effective mechanism for tracking changes and updating the model for future developments and use cases. The Proposed ontology can be used for AFDD model development, tracking changes, analytics, visualization, and digital twinning by enabling integration of BIM with BAS/BMS (Building Automation System/Building Management System) concepts and secondly to store AFDD configuration and analytics in the AFDDOnto. Select competency questions are constructed using SPARQL queries to access the proposed knowledge model. The proposed ontology has been tested against different measures using multiple metrics and a case study and further validated using a semi-structured survey of experts. Applied AI engineers, Facility managers, Asset managers, and building owners aiming to develop AFDD models for HVAC systems can benefit from adopting this ontology for HVAC maintenance, including analysis, model development, and knowledge management.

Career Insight: Search Professional

Abstract Amey Dharwadker, AI Engineering Manager at Meta and winner of BCS Search Professional of the Year 2023, shares career insights with Grant Powell MBCS.

The Influence of AI in the Media Work Force: How Companies Use an Array of Legal Remedies

The emergence of ChatGPT and other related artificial intelligence systems has posed many questions upon the impact that such tools could have on some jobs, including media workers. Serious legal concerns have arisen regarding the learning practices of AI-related companies such as OpenAI and Google. These concerns involve crawling and extracting presumably unauthorized copyright works from news repositories, whose rightholders are often media companies. In this article, we aim to categorize the newsroom practices and routines affected by artificial intelligence. We also explore copyright-law related issues, including AI-assisted reporting, its impact on journalists and the media workforce, SEO and commercial strategies, as well as training and blocking AI engines. The legal solutions applied to solve those questions are also addressed, including technical solutions, fair use guidelines and legal solutions (litigation, legislative reform, and negotiation). Our conclusion is twofold: first, in the unequal fight against artificial intelligence systems, a utilitarian and entrepreneurial conception of intellectual property is enforced; and second, the position of journalists as authors is weakening.

Advancements in Reinforcement Learning

Reinforcement Learning (RL) stands at the forefront of machine learning, offering a paradigm where agents learn to navigate complex environments through trial and error interactions. This paper provides a comprehensive overview of RL concepts, algorithms, challenges, and future directions. Key concepts such as agent-environment interaction, Markov Decision Processes (MDPs), value-based and policy-based methods are elucidated. Challenges including the exploration-exploitation trade-off, sample efficiency, and safety concerns are discussed, alongside potential breakthroughs in deep RL, handling continuous action spaces, and dealing with partial observability. The integration of RL with emerging technologies like IoT and its implications for real-world applications such as robotics, autonomous vehicles, and healthcare are explored. Finally, future trends and directions in RL research, including AI engineering and automated feature engineering, are outlined, highlighting the potential for transformative advancements and their impact on various domains. This paper serves as a comprehensive resource for researchers, practitioners, and enthusiasts interested in the evolving landscape of Reinforcement Learning

Artificial-Neural-Network-Based Predicted Model for Seam Strength of Five-Pocket Denim Jeans: A Review

This study explores previous research efforts concerning prediction models related to the textile and polymer industry, especially garment seam strength, emphasizing critical parameters such as stitch density, fabric GSM, thread type, thread count, stitch classes, and seam types. These parameters play a pivotal role in determining the durability and overall quality of denim jeans based on cellulosic polymer. A significant focus is dedicated to the mathematical computational models employed for predicting seam strength in five-pocket denim jeans. Herein, the discussion poses the application of AI for manufacturing industries, especially for textile and clothing sectors, and highlights the importance of using a machine learning prediction model for sewing thread consumption, seam strength analysis, and seam performance analysis. Therefore, the authors suggest the significant importance of the machine learning prediction model, as future trends anticipate advancements in AI-driven methodologies, potentially leading to high-profile predictions and superior manufacturing processes. The authors also describe the limitation of AI and address a comprehensive model of risk outlines of AI in the manufacturing-based industries, especially the garments industry. Put simply, this review serves as a bridge between the realms of AI, mathematics, and textile engineering, providing a clear understanding of how artificial-neural-network-based models will be shaping the future of seam strength prediction in the denim manufacturing landscape. This type of evolution, based on ANN, will support and enhance the accuracy and efficiency of seam strength predictions by allowing models to discern intricate patterns and relationships within vast and diverse datasets.

Chess using AI Review

This research paper explores the development of an AI chess engine leveraging the Minimax algorithm with Alpha-Beta pruning technique, implemented using Python and JavaScript programming languages. The objective is to investigate the effectiveness of these algorithms in creating a proficient and competitive chess-playing AI. The paper begins with an overview of the Minimax algorithm and its application in game theory, followed by an explanation of Alpha-Beta pruning and its role in enhancing the efficiency of the search process within the game tree. Subsequently, the implementation details of the AI chess engine in both Python and JavaScript are discussed, highlighting the design considerations, algorithmic optimizations, and programming techniques utilized. The paper also provides insights into the integration of the AI engine with a graphical user interface (GUI) for interactive gameplay experiences. Additionally, experimental results and performance evaluations are presented to assess the AI engine's strength and playing capabilities against human players or other AI opponents

Responsible AI Pattern Catalogue: A Collection of Best Practices for AI Governance and Engineering

Responsible Artificial Intelligence (RAI) is widely considered as one of the greatest scientific challenges of our time and is key to increase the adoption of Artificial Intelligence (AI). Recently, a number of AI ethics principles frameworks have been published. However, without further guidance on best practices, practitioners are left with nothing much beyond truisms. In addition, significant efforts have been placed at algorithm level rather than system level, mainly focusing on a subset of mathematics-amenable ethical principles, such as fairness. Nevertheless, ethical issues can arise at any step of the development lifecycle, cutting across many AI and non-AI components of systems beyond AI algorithms and models. To operationalize RAI from a system perspective, in this article, we present an RAI Pattern Catalogue based on the results of a multivocal literature review. Rather than staying at the principle or algorithm level, we focus on patterns that AI system stakeholders can undertake in practice to ensure that the developed AI systems are responsible throughout the entire governance and engineering lifecycle. The RAI Pattern Catalogue classifies the patterns into three groups: multi-level governance patterns, trustworthy process patterns, and RAI-by-design product patterns. These patterns provide systematic and actionable guidance for stakeholders to implement RAI.

Leveraging the Synergy of IPv6, Generative AI, and Web Engineering to Create a Big Data-driven Education Platform

The rapid advancement of network technology in China has significantly accelerated the implementation of information technology in higher education. Through the utilization of computer technology, multimedia technology, big data technology, artificial intelligence technology, and network communication technology, the integration of these technologies in university teaching has become widespread. This paper presents an analysis and discussion on the utilization of the latest IPv6 network transmission protocol technology to enhance the application of data collection in university education, with a specific focus on gathering information related to university faculties. By leveraging web engineering and multimedia technology as fundamental components, the network facilitates the sharing of educational resources among students, thereby enabling the reform of management approaches, fostering educational progress in China, and establishing a comprehensive big data-driven education platform specifically tailored to colleges and universities. Additionally, the incorporation of big data visualization and analysis tools allows for easy retrieval of existing university educational information, facilitates the creation of data charts, and expedites the utilization of data for its inherent value. Finally, the proposed approach employs generative AI to collect and analyze feedback from students and educators, followed by the application of web engineering techniques to continuously enhance the online education platform based on this feedback.

IoHT Based Tele-Healthcare Support System for Feasibility and Perfor-mance Analysis

With the rapid development in Internet of Things (IoT) it come up with the subbranch Internet of Healthcare Things or the Internet of Medical Things. With the help of portable healthcare sensors and the wearable devices now, ambulance can be converted into critical diagnostics unit on the wheels. Monitoring the home care patient is not the just part of IoT application where doctor can remotely monitor the patient situation by the means of sensors. It requires a complete system may called as telemedicine where multiple components will play similar priority role to monitor the patient status using sensors, let the doctor connect the patient, data analytics calculate the progression, AI engine will detect the emergency and act upon. The proposed system is to design and developing centralized tele-healthcare support system and systematic analytics sharing with multiple level of user with privacy protection. This will allow different authorities to take advantages from patient generated records.

XVDPU: A High-Performance CNN Accelerator on the Versal Platform Powered by the AI Engine

Today, convolutional neural networks (CNNs) are widely used in computer vision applications. However, the trends of higher accuracy and higher resolution generate larger networks. The requirements of computation or I/O are the key bottlenecks. In this article, we propose XVDPU: the AI Engine (AIE)-based CNN accelerator on Versal chips to meet heavy computation requirements. To resolve the IO bottleneck, we adopt several techniques to improve data reuse and reduce I/O requirements. An arithmetic logic unit is further proposed that can better balance resource utilization, new feature support, and efficiency of the whole system. We have successfully deployed more than 100 CNN models with our accelerator. Our experimental results show that the 96-AIE-core implementation can achieve 1,653 frames per second (FPS) for ResNet50 on VCK190, which is 9.8× faster than the design on ZCU102 running at 168.5 FPS. The 256-AIE-core implementation can further achieve 4,050 FPS. We propose a tilling strategy to achieve feature-map-stationary for high-definition CNN with the accelerator, achieving 3.8× FPS improvement on the residual channel attention network and 3.1× on super-efficient super-resolution. This accelerator can also solve the 3D convolution task in disparity estimation, achieving end-to-end performance of 10.1 FPS with all the optimizations.