國立高雄大學圖資館 |

登入

回首頁

Beyond Misinformation: Understanding AI Hallucinations in Science Communication /

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Beyond Misinformation: Understanding AI Hallucinations in Science Communication /Anqi Shao.
作者:	Shao, Anqi,
面頁冊數:	1 electronic resource (165 pages)
附註:	Source: Dissertations Abstracts International, Volume: 87-03, Section: B.
附註:	Advisors: Scheufele, Dietram Committee members: Brossard, Dominique; Chen, Kaipin; Claessens, Amy; Li, Sharon.
Contained By:	Dissertations Abstracts International87-03B.
標題:	Web studies.
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=32240134
ISBN:	9798293850983

Beyond Misinformation: Understanding AI Hallucinations in Science Communication /
Shao, Anqi,

Beyond Misinformation: Understanding AI Hallucinations in Science Communication /Anqi Shao. - 1 electronic resource (165 pages)

Source: Dissertations Abstracts International, Volume: 87-03, Section: B.

This dissertation examines hallucination from artificial intelligence (AI) systems as a new source of inaccuracy in science communication. In the first chapter, I distinguished AI hallucination as different from human misinformation and proposed a supply-demand framework for studying how these errors are produced, interpreted, and acted upon. In the following empirical chapters, Chapter 2 audits current popular large language models (LLMs) and found scientific prompts with different levels uncertainty and controversy are responded differently. Chapter 2 also catalogs hallucination types including the most frequent citation fabrication and misrepresentation of the cited sources. Chapter 3 used a national survey to segment U.S. publics into the Informed, Skeptics, and Observers in terms of their perceptions of AI hallucination and support for regulations and found concern about hallucinations did not always relate to awareness of hallucinations and was shaped by beliefs and predispositions. Chapter 4 experimented on the immediate belief updates when users were exposed to hallucinations with neutral prompt and inaccurate response, in comparison with other conditions. The sensitivity for the belief update varied among different literacy groups, and the updated beliefs shifted subsequent choices for further information seeking. Together, this dissertation provided a brief and preliminary map of where and how AI hallucinations may arise in science communication, who worries about them, and how the users may react to them, with suggestions for science communication practice, interface designs, and governance.

English

ISBN: 9798293850983Subjects--Topical Terms:

708690
Web studies.
Subjects--Index Terms:

Large language models

Beyond Misinformation: Understanding AI Hallucinations in Science Communication /
LDR:03150nmm a22004333i 4500 001 693759
005 20260511063021.5
006 m o d
007 cr|nu||||||||
008 260819s2025 miu||||||m |||||||eng d
020 $a 9798293850983
035 $a (MiAaPQD)AAI32240134
035 $a AAI32240134
040 $a MiAaPQD $b eng $c MiAaPQD $e rda
100 1 $a Shao, Anqi, $e author. $3 1010421
245 1 0 $a Beyond Misinformation: Understanding AI Hallucinations in Science Communication / $c Anqi Shao.
264 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2025
300 $a 1 electronic resource (165 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: 87-03, Section: B.
500 $a Advisors: Scheufele, Dietram Committee members: Brossard, Dominique; Chen, Kaipin; Claessens, Amy; Li, Sharon.
502 $b Ph.D. $c The University of Wisconsin - Madison $d 2025.
520 $a This dissertation examines hallucination from artificial intelligence (AI) systems as a new source of inaccuracy in science communication. In the first chapter, I distinguished AI hallucination as different from human misinformation and proposed a supply-demand framework for studying how these errors are produced, interpreted, and acted upon. In the following empirical chapters, Chapter 2 audits current popular large language models (LLMs) and found scientific prompts with different levels uncertainty and controversy are responded differently. Chapter 2 also catalogs hallucination types including the most frequent citation fabrication and misrepresentation of the cited sources. Chapter 3 used a national survey to segment U.S. publics into the Informed, Skeptics, and Observers in terms of their perceptions of AI hallucination and support for regulations and found concern about hallucinations did not always relate to awareness of hallucinations and was shaped by beliefs and predispositions. Chapter 4 experimented on the immediate belief updates when users were exposed to hallucinations with neutral prompt and inaccurate response, in comparison with other conditions. The sensitivity for the belief update varied among different literacy groups, and the updated beliefs shifted subsequent choices for further information seeking. Together, this dissertation provided a brief and preliminary map of where and how AI hallucinations may arise in science communication, who worries about them, and how the users may react to them, with suggestions for science communication practice, interface designs, and governance.
546 $a English
590 $a School code: 0262
650 4 $a Web studies. $3 708690
650 4 $a Computer science. $3 199325
653 $a Large language models
653 $a Science communication
653 $a Skeptics
653 $a Hallucination types
653 $a Human misinformation
690 $a 0800
690 $a 0646
690 $a 0984
710 2 $a The University of Wisconsin - Madison. $b Mass Communications - AG. $e degree granting institution. $3 1010422
720 1 $a Scheufele, Dietram $e degree supervisor.
773 0 $t Dissertations Abstracts International $g 87-03B.
790 $a 0262
791 $a Ph.D.
792 $a 2025
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=32240134