東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁 到查詢結果 [ null ]

切換: 標籤 | MARC模式 | ISBD

Stakeholder-Centered Challenges in R...

Bell, Andrew Loren,

FindBook

Google Book

Amazon

博客來

Stakeholder-Centered Challenges in Responsible Artificial Intelligence /

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Stakeholder-Centered Challenges in Responsible Artificial Intelligence // Andrew Loren Bell.
作者:	Bell, Andrew Loren,
面頁冊數:	1 electronic resource (214 pages)
附註:	Source: Dissertations Abstracts International, Volume: 86-11, Section: B.
Contained By:	Dissertations Abstracts International86-11B.
標題:	Computer science. -
電子資源:	https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=32039992
ISBN:	9798315745433

Stakeholder-Centered Challenges in Responsible Artificial Intelligence /
Bell, Andrew Loren,

Stakeholder-Centered Challenges in Responsible Artificial Intelligence /Andrew Loren Bell. - 1 electronic resource (214 pages)

Source: Dissertations Abstracts International, Volume: 86-11, Section: B.

Recent years have seen rapid progress and the widespread adoption of artificial intelligence (AI) and machine learning (ML) systems into the public and private sectors. This dissertation explores AI systems in socio-technical contexts, highlighting the need to move beyond purely technical perspectives and consider human stakeholders. The recognition of risks and harms associated with AI systems, particularly for marginalized and vulnerable groups, has given rise to the field of Responsible Artificial Intelligence (RAI). This work focuses on stakeholder-centered practical challenges in RAI, and presents findings from seven research projects that fall broadly into three main categories: trade-offs, algorithmic recourse, and advocacy.In the area of trade-offs, we investigate the complex relationship between accuracy and explainability, challenge the "Impossibility Theorem" in algorithmic fairness by demonstrating how hard limits can be circumvented by relaxing constraints, and introduce SafeNudge, a Large Language Model safeguard with tunable safety-performance trade-offs. Importantly, all these trade-offs involve navigating multiple stakeholder objectives, like the utility of an AI system versus it's adherence to human values. Two works are discussed in algorithmic recourse: the first examines the criticality of considering temporal dynamics when implementing algorithmic recourse in practice, and the second explores unfairness in recourse that can arise due to group-wise differences in capacity for effort. Finally, in RAI advocacy, this work investigates how to transfer learnings from the research setting into the real-world. We present a stakeholder-first approach for designing regulatory-compliant AI systems, and an approach for creating "transparency advocates" who can drive change within organizations. Overall, this dissertation contributes to the field of RAI by providing practical insights and tools for navigating the complexities of developing, deploying, and governing AI systems through a stakeholder-centered lens.

English

ISBN: 9798315745433Subjects--Topical Terms:

523869
Computer science.
Subjects--Index Terms:

AI safety

Stakeholder-Centered Challenges in Responsible Artificial Intelligence /
LDR:03643nmm a22004693i 4500 001 2434608
005 20250912102051.5
006 m o d
007 cr|nu||||||||
008 261011s2025 miu||||||m |||||||eng d
020 $a 9798315745433
035 $a (MiAaPQD)AAI32039992
035 $a AAI32039992
040 $a MiAaPQD $b eng $c MiAaPQD $e rda
100 1 $a Bell, Andrew Loren, $e author. $3 3824640
245 1 0 $a Stakeholder-Centered Challenges in Responsible Artificial Intelligence / $c Andrew Loren Bell.
264 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2025
300 $a 1 electronic resource (214 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: 86-11, Section: B.
500 $a Advisors: Stoyanovich, Julia; Nov, Oded Committee members: Ghani, Rayid; Friedler, Sorelle.
502 $b Ph.D. $c New York University Tandon School of Engineering $d 2025.
520 $a Recent years have seen rapid progress and the widespread adoption of artificial intelligence (AI) and machine learning (ML) systems into the public and private sectors. This dissertation explores AI systems in socio-technical contexts, highlighting the need to move beyond purely technical perspectives and consider human stakeholders. The recognition of risks and harms associated with AI systems, particularly for marginalized and vulnerable groups, has given rise to the field of Responsible Artificial Intelligence (RAI). This work focuses on stakeholder-centered practical challenges in RAI, and presents findings from seven research projects that fall broadly into three main categories: trade-offs, algorithmic recourse, and advocacy.In the area of trade-offs, we investigate the complex relationship between accuracy and explainability, challenge the "Impossibility Theorem" in algorithmic fairness by demonstrating how hard limits can be circumvented by relaxing constraints, and introduce SafeNudge, a Large Language Model safeguard with tunable safety-performance trade-offs. Importantly, all these trade-offs involve navigating multiple stakeholder objectives, like the utility of an AI system versus it's adherence to human values. Two works are discussed in algorithmic recourse: the first examines the criticality of considering temporal dynamics when implementing algorithmic recourse in practice, and the second explores unfairness in recourse that can arise due to group-wise differences in capacity for effort. Finally, in RAI advocacy, this work investigates how to transfer learnings from the research setting into the real-world. We present a stakeholder-first approach for designing regulatory-compliant AI systems, and an approach for creating "transparency advocates" who can drive change within organizations. Overall, this dissertation contributes to the field of RAI by providing practical insights and tools for navigating the complexities of developing, deploying, and governing AI systems through a stakeholder-centered lens.
546 $a English
590 $a School code: 1988
650 4 $a Computer science. $3 523869
650 4 $a Systems science. $3 3168411
650 4 $a Robotics. $3 519753
653 $a AI safety
653 $a Algorithmic fairness
653 $a Explainability
653 $a Machine learning
653 $a Algorithmic recourse
653 $a Responsible Artificial Intelligence
690 $a 0800
690 $a 0984
690 $a 0771
690 $a 0790
710 2 $a New York University Tandon School of Engineering. $b Computer Science & Engineering. $e degree granting institution. $3 3817126
720 1 $a Stoyanovich, Julia $e degree supervisor.
720 1 $a Nov, Oded $e degree supervisor.
773 0 $t Dissertations Abstracts International $g 86-11B.
790 $a 1988
791 $a Ph.D.
792 $a 2025
856 4 0 $u https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=32039992