國立虎尾科技大學 |

Human-AI Collaboration in Software Development: A Multi-Method Investigation of Vulnerability Introduction Prediction and Code Generation /

紀錄類型:	書目-語言資料,印刷品 : Monograph/item
正題名/作者:	Human-AI Collaboration in Software Development: A Multi-Method Investigation of Vulnerability Introduction Prediction and Code Generation // Agrim Sachdeva.
作者:	Sachdeva, Agrim,
面頁冊數:	1 electronic resource (104 pages)
附註:	Source: Dissertations Abstracts International, Volume: 86-01, Section: B.
Contained By:	Dissertations Abstracts International86-01B.
標題:	Information science. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=31327907
ISBN:	9798383188200

Human-AI Collaboration in Software Development: A Multi-Method Investigation of Vulnerability Introduction Prediction and Code Generation /
Sachdeva, Agrim,

Human-AI Collaboration in Software Development: A Multi-Method Investigation of Vulnerability Introduction Prediction and Code Generation /Agrim Sachdeva. - 1 electronic resource (104 pages)

Source: Dissertations Abstracts International, Volume: 86-01, Section: B.

Artificial intelligence (AI)-based technologies, especially predictive analytics, and generative AI are radically changing organizations and processes. AI technologies complement human abilities and skills to improve the resulting intelligence of the human-AI ensemble. This dissertation investigates the complementary roles of humans and AI in software development, maintenance, and security, specifically two critical facets of human-AI collaboration in the application domain of software development: vulnerability introduction prediction in open-source software (OSS) and code generation.Essay 1 examines the role of AI in OSS security, as OSS underpins much of the modern digital infrastructure and is contributed to by many known developers worldwide. Developer training is an important vulnerability management strategy. However, generalized training leads to training fatigue and suboptimal allocation of organizational resources. Therefore, a dynamic graph representation learning-based deep learning framework is proposed, wherein vulnerability introduction is proactively predicted. The predictive model can serve as input to organizational decision-makers for conducting personalized and proactive training. This proposed human-AI collaboration underpins the design and evaluation of the proposed framework, situated within the computational design science paradigm. The study contributes by evaluating the framework against prevailing methods and contributing two general design principles.While the first essay focuses on the security of OSS, the second essay focuses on the process of code generation using Large Language Models (LLMs). AI's potential to automate the generation of software components can not only increase efficiency but also allow human developers to focus more on strategic, high-level aspects of software development. Situated within the behavioral research paradigm, the second essay proposes a lab experiment to determine the effects of using LLMs and examine the underlying mechanisms and boundary conditions for the same. The expected theoretical contribution for essay two is to literature on the autonomy and adoption of AI agents in a programming context.In conclusion, the dissertation studies the transformative potential of human-AI collaboration in the context of software development and security. Essay 1, rooted in the computational design science paradigm, focuses on how AI can enhance open-source software security by predicting the introduction of vulnerabilities for proactive vulnerability management. Essay 2, situated in the behavioral research paradigm, explores the unintended effects of the automation of code generation with LLMs. Together, the two essays contribute to IS theory and practice by examining the emerging phenomena resulting from human-AI collaboration in software development.

English

ISBN: 9798383188200Subjects--Topical Terms:

561178
Information science.
Subjects--Index Terms:

Cybersecurity

Human-AI Collaboration in Software Development: A Multi-Method Investigation of Vulnerability Introduction Prediction and Code Generation /
LDR:04387nam a22004333i 4500 001 1157813
005 20250603111420.5
006 m o d
007 cr|nu||||||||
008 250804s2024 miu||||||m |||||||eng d
020 $a 9798383188200
035 $a (MiAaPQD)AAI31327907
035 $a AAI31327907
040 $a MiAaPQD $b eng $c MiAaPQD $e rda
100 1 $a Sachdeva, Agrim, $e author. $3 1484088
245 1 0 $a Human-AI Collaboration in Software Development: A Multi-Method Investigation of Vulnerability Introduction Prediction and Code Generation / $c Agrim Sachdeva.
264 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2024
300 $a 1 electronic resource (104 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 86-01, Section: B.
500 $a Advisors: Dennis, Alan R. Committee members: Samtani, Sagar; Kim, Antino; Boyle, Ernest O'.
502 $b Ph.D. $c Indiana University $d 2024.
520 $a Artificial intelligence (AI)-based technologies, especially predictive analytics, and generative AI are radically changing organizations and processes. AI technologies complement human abilities and skills to improve the resulting intelligence of the human-AI ensemble. This dissertation investigates the complementary roles of humans and AI in software development, maintenance, and security, specifically two critical facets of human-AI collaboration in the application domain of software development: vulnerability introduction prediction in open-source software (OSS) and code generation.Essay 1 examines the role of AI in OSS security, as OSS underpins much of the modern digital infrastructure and is contributed to by many known developers worldwide. Developer training is an important vulnerability management strategy. However, generalized training leads to training fatigue and suboptimal allocation of organizational resources. Therefore, a dynamic graph representation learning-based deep learning framework is proposed, wherein vulnerability introduction is proactively predicted. The predictive model can serve as input to organizational decision-makers for conducting personalized and proactive training. This proposed human-AI collaboration underpins the design and evaluation of the proposed framework, situated within the computational design science paradigm. The study contributes by evaluating the framework against prevailing methods and contributing two general design principles.While the first essay focuses on the security of OSS, the second essay focuses on the process of code generation using Large Language Models (LLMs). AI's potential to automate the generation of software components can not only increase efficiency but also allow human developers to focus more on strategic, high-level aspects of software development. Situated within the behavioral research paradigm, the second essay proposes a lab experiment to determine the effects of using LLMs and examine the underlying mechanisms and boundary conditions for the same. The expected theoretical contribution for essay two is to literature on the autonomy and adoption of AI agents in a programming context.In conclusion, the dissertation studies the transformative potential of human-AI collaboration in the context of software development and security. Essay 1, rooted in the computational design science paradigm, focuses on how AI can enhance open-source software security by predicting the introduction of vulnerabilities for proactive vulnerability management. Essay 2, situated in the behavioral research paradigm, explores the unintended effects of the automation of code generation with LLMs. Together, the two essays contribute to IS theory and practice by examining the emerging phenomena resulting from human-AI collaboration in software development.
546 $a English
590 $a School code: 0093
650 4 $a Information science. $3 561178
650 4 $a Computer science. $3 573171
653 $a Cybersecurity
653 $a Large language models
653 $a Open-source software
653 $a Psychological ownership
690 $a 0310
690 $a 0984
690 $a 0723
690 $a 0800
710 2 $a Indiana University. $b Business. $3 1179133
720 1 $a Dennis, Alan R. $e degree supervisor.
773 0 $t Dissertations Abstracts International $g 86-01B.
790 $a 0093
791 $a Ph.D.
792 $a 2024
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=31327907