Building a construction law knowledge repository to enhance general-purpose large language model performance on domain question-answering: a case of China DOI
Shenghua Zhou, Hongyu Wang, S. Thomas Ng

et al.

Engineering Construction & Architectural Management, Journal Year: 2025, Volume and Issue: unknown

Published: April 30, 2025

Purpose Achieving smart question-answering (QA) for construction laws (CLs) holds significant promise in aiding domain professionals with legal inquiries. Existing studies of law (CLQA) rely on learning-based models, which require extensive training data and are limited to a narrow QA scope. Meanwhile, general-purpose large language models (GPLLMs) possess great potential CLQA but fall short domain-specific knowledge. This study aims propose data-driven expertise-based approach develop knowledge repository (CLKR) validate its effectiveness enhancing the performance GPLLMs. Design/methodology/approach methodology includes (1) recognizing 702 candidate CL documents from 374,992 official judgments, (2) building CLKR 387 filtered covering eight areas, (3) integrating seven representative GPLLMs (4) constructing 2,140-question dataset Professional Construction Engineer Qualification Examinations (PCEQEs) during 2014–2023 compare between pairs without CLKR. Findings The significantly enhances GPLLMs, yielding an impressive average accuracy increase 21.1%, individual improvements ranging 9.9 44.9%. Furthermore, boosts single-answer questions by 14.9% multiple-answer 38.3%. Additionally, enhancements across 8 areas 14.5 28.2%. Originality/value proposes developing external base empower expanding scope while bypassing complex traditional models. Moreover, this confirms augmenting GPLLM offers reusable test as benchmark.

Language: Английский

Perception, performance, and detectability of conversational artificial intelligence across 32 university courses DOI Creative Commons
Hazem Ibrahim, Fengyuan Liu,

Rohail Asim

et al.

Scientific Reports, Journal Year: 2023, Volume and Issue: 13(1)

Published: Aug. 24, 2023

Language: Английский

Citations

93
Artificial Intelligence-Enabled Intelligent Assistant for Personalized and Adaptive Learning in Higher Education DOI Creative Commons
Ramteja Sajja, Yusuf Sermet,

Muhammed Cikmaz

et al.

Information, Journal Year: 2024, Volume and Issue: 15(10), P. 596 - 596

Published: Sept. 30, 2024

This paper presents a novel framework, artificial intelligence-enabled intelligent assistant (AIIA), for personalized and adaptive learning in higher education. The AIIA system leverages advanced AI natural language processing (NLP) techniques to create an interactive engaging platform. platform is engineered reduce cognitive load on learners by providing easy access information, facilitating knowledge assessment, delivering support tailored individual needs styles. AIIA’s capabilities include understanding responding student inquiries, generating quizzes flashcards, offering pathways. research findings have the potential significantly impact design, implementation, evaluation of AI-enabled virtual teaching assistants (VTAs) education, informing development innovative educational tools that can enhance outcomes, engagement, satisfaction. methodology, architecture, services, integration with management systems (LMSs) while discussing challenges, limitations, future directions

Language: Английский

Citations

89
What drives tourists’ continuance intention to use ChatGPT for travel services? A stimulus-organism-response perspective DOI

Hong Chuong Pham,

Cong Doanh Duong,

Giang Khanh Huyen Nguyen

et al.

Journal of Retailing and Consumer Services, Journal Year: 2024, Volume and Issue: 78, P. 103758 - 103758

Published: Feb. 9, 2024

Language: Английский

Citations

62
Enhancing Mathematical Capabilities through ChatGPT and Similar Generative Artificial Intelligence: Roles and Challenges in Solving Mathematical Problems DOI

Nitin Rane

SSRN Electronic Journal, Journal Year: 2023, Volume and Issue: unknown

Published: Jan. 1, 2023

The incorporation of artificial intelligence (AI) technologies, with a particular focus on generative models such as ChatGPT, has ushered in revolutionary era the field education, fundamentally altering way students engage mathematical problem-solving. This scholarly article investigates diverse roles and obstacles associated harnessing potential ChatGPT other AI tools to enhance proficiency. By implementing these cutting-edge educators can provide tailored learning experiences that cater wide array styles paces. enables receive immediate feedback, participate interactive problem-solving dialogues, access step-by-step solutions align their individual requirements. Nevertheless, integration into education is not without its complexities. A central challenge revolves around ensuring accuracy dependability AI-generated solutions, mitigating any biases present training data, maintaining harmonious balance between automation human intervention process. Ethical concerns, encompassing matters data privacy ethical use AI, also demand meticulous consideration. Furthermore, delves cognitive impacts students, scrutinizing how reliance might affect critical thinking skills grasp fundamental concepts. It explores methodologies aimed at fostering synergistic relationship intelligence, encouraging development robust strategies while computational capabilities AI. research illuminates promising incorporating notably realm mathematics education. addressing challenges adopting nuanced approach, harness power craft enriching efficient environments, nurturing generation individuals adept well-equipped tackle intricacies modern world.

Language: Английский

Citations

47
ChatGPT performance on multiple choice question examinations in higher education. A pragmatic scoping review DOI Creative Commons
Philip M. Newton, Maira Xiromeriti

Assessment & Evaluation in Higher Education, Journal Year: 2024, Volume and Issue: 49(6), P. 781 - 798

Published: Jan. 17, 2024

Media coverage suggests that ChatGPT can pass examinations based on multiple choice questions (MCQs), including those used to qualify doctors, lawyers, scientists etc. This poses a potential risk the integrity of examinations. We reviewed current research evidence regarding performance MCQ-based in higher education, along with recommendations for how educators might address challenges and benefits arising from these data. 53 studies were included, covering 114 question sets, totalling 49014 MCQs. Free versions upon GPT-3/3.5 performed better than random guessing but failed most examinations, performing significantly worse average human student. GPT-4 passed was par subjects. These findings indicate all summative assessments should be conducted under secure conditions restricted access similar tools, particularly which assess foundational knowledge.

Language: Английский

Citations

27
A Cross-National Assessment of Artificial Intelligence (AI) Chatbot User Perceptions in Collegiate Physics Education. DOI Creative Commons

Benjamin Osafo Agyare,

Joseph Asare, A. F. Kraishan

et al.

Computers and Education Artificial Intelligence, Journal Year: 2025, Volume and Issue: unknown, P. 100365 - 100365

Published: Jan. 1, 2025

Language: Английский

Citations

2
Professional Certification Benchmark Dataset: The First 500 Jobs for Large Language Models DOI Open Access

David Noever,

Matt Ciolino

Published: July 22, 2023

The research creates a professional certification survey to test large language models and evaluate their employable skills. It compares the performance of two AI models, GPT-3 Turbo-GPT3.5, on benchmark dataset 1149 certifications, emphasizing vocational readiness rather than academic performance. achieved passing score (>70% correct) in 39% certifications without fine-tuning or exam preparation. demonstrated qualifications various computer-related fields, such as cloud virtualization, business analytics, cybersecurity, network setup repair, data analytics. Turbo-GPT3.5 scored 100% valuable Offensive Security Certified Professional (OSCP) exam. also displayed competence other domains, including nursing, licensed counseling, pharmacy, teaching. passed Financial Industry Regulatory Authority (FINRA) Series 6 with 70% grade Interestingly, performed well customer service tasks, suggesting potential applications human augmentation for chatbots call centers routine advice services. sensory experience-based tests wine sommelier, beer taster, emotional quotient, body reader. OpenAI model improvement from Babbage Turbo resulted median 60% better-graded less few years. This progress suggests that focusing latest model's shortcomings could lead highly performant capable mastering most demanding certifications. We open-source expand range testable skills improve gain emergent capabilities.

Language: Английский

Citations

38
Accurately detecting AI text when ChatGPT is told to write like a chemist DOI Creative Commons
Heather Desaire, Aleesa E. Chua, Min Kim

et al.

Cell Reports Physical Science, Journal Year: 2023, Volume and Issue: 4(11), P. 101672 - 101672

Published: Nov. 1, 2023

Large language models like ChatGPT can generate authentic-seeming text at lightning speed, but many journal publishers reject as authors on manuscripts. Thus, a means to accurately distinguish human-generated from artificial intelligence (AI)-generated is immediately needed. We recently developed an accurate AI detector for scientific journals and, herein, test its ability in variety of challenging situations, including human wide chemistry journals, the most advanced publicly available model (GPT-4), important, generated using prompts designed obfuscate use. In all cases, and was assigned with high accuracy. ChatGPT-generated be readily detected journals; this advance fundamental prerequisite understanding how automated generation will impact publishing now into future.

Language: Английский

Citations

27
Exploring the proficiency of ChatGPT-4: An evaluation of its performance in the Taiwan advanced medical licensing examination DOI Creative Commons
Hong-Mo Shih,

Pak Ki Chan,

Wu‐Huei Hsu

et al.

Digital Health, Journal Year: 2024, Volume and Issue: 10

Published: Jan. 1, 2024

Taiwan is well-known for its quality healthcare system. The country's medical licensing exams offer a way to evaluate ChatGPT's proficiency.

Language: Английский

Citations

13
Generative AI chatbots in higher education: a review of an emerging research area DOI Creative Commons
Cormac McGrath, Alexandra Farazouli, Teresa Cerratto Pargman

et al.

Higher Education, Journal Year: 2024, Volume and Issue: unknown

Published: Aug. 24, 2024

Abstract Artificial intelligence (AI) chatbots trained on large language models are an example of generative AI which brings promises and threats to the higher education sector. In this study, we examine emerging research area in (HE), focusing specifically empirical studies conducted since release ChatGPT. Our review includes 23 articles published between December 2022 2023 exploring use HE settings. We take a three-pronged approach data. first state field HE. Second, identify theories learning used Third, scrutinise discourses framing latest work chatbots. findings contribute better understanding eclectic nascent HE, lack common conceptual groundings about human learning, presence both dystopian utopian future role

Language: Английский

Citations

11