摘 要：近年来，人工智能技术驱动新闻业向智能化转向，人工智能技术在新闻领域的应用，呈现出由边缘化（终端、设备）向中心化（新闻生产）突破的特征。生产主体作为新闻生产实践活动中的重要构成要素，起着直接发起者、直接参与者、直接呈现者的关键作用。无论哪种媒体形态，播音员主持人始终站在新闻生产与信息传播的最前沿，直接面对受众，成为决定传播效果的重要角色。人工智能主播作为智能传播媒介表征符号的“代言人”，是智能生产平台“算法把关”下的技术主体，是新型主流媒体在人工智能技术加持下媒介深度融合发展的必然产物。人类传播技术的每一次革新，都会带来媒介形态的变化、人与信息间关系的变化、获取与反馈信息方式的变化以及一切媒介作为人的延伸，视野与知觉的变化。
本文的第一章节，在明确所选主题具有重要理论意义与社会意义的基础上，在对现阶段已有的相关研究的全面剖析与总结中，提出本文的研究问题：受众对于人工智能主播播报新闻的认知效果与情感体验。包括在受众“接码——解码——收码”、“认识——情感——意志”的过程中，人工智能主播作为传播主体的出现，是否正向影响了受众“收码”的结果？是什么样的因素和指标影响了受众的“收码”？创作主体在传播中的自然属性规律与社会属性规律的改变对传播效果产生了何种影响？  
本文的第二章节，首先对人工智能主播的类别及特征进行划分，明确本文研究对象为我国人工智能新闻主播。其次，通过梳理人工智能主播的发展脉络、技术特点、应用模式和发展趋势，挖掘不同时期人工智能主播出现的社会背景与技术应用。再次，选取典型案例探究不同社会背景下人工智能主播诞生、发展与退出历史舞台背后的社会动因。
本文的第三章和第四章，探究受众对现阶段人工智能主播播报新闻的感知、评价与记忆效果。其中，第三章节围绕实验一展开。运用在线实验法，采用自主录制、自主制作的真人主播与人工智能主播播报新闻的视频作为实验材料，进行了一项2（主播类别：人工智能主播vs.真人主播）×2（新闻类型：民生新闻vs.时政新闻）×2（平台属性：官方媒体vs.自媒体）的在线实验。结果表明，主播类别、新闻类型和平台属性均显著影响受众的感知与评价。人工智能主播播报的新闻在可信度、新闻代入感、受众喜爱度、整体质量评价方面均低于真人主播。此外，被试对民生新闻与时政新闻的整体质量评价并无显著差异，但民生新闻的可信度、代入感均显著高于时政新闻且更受被试喜欢；无论官方媒体还是自媒体发布的视频，整体质量评价无显著差异，然而，带有官方媒体标识的视频被认为更可信、更有代入感并且更受被试喜欢。 
本文的第四章节，围绕实验二展开。研究受众观看人工智能主播播报新闻的记忆效果以及与观看真人主播播报新闻的记忆效果差异。同样运用控制实验法，将相同内容的新闻文稿由人工智能主播与真人主播分别进行播报并录制，基于社会化线索与传者身份的双重视角。其中，在社会化线索设置方面，为真人主播设计体态系统与声音副语言系统的社会化线索，采用设置社会化线索视频与未设置社会化线索视频对比的方式检验社会化线索的作用；另通过问卷调查的方法考察传播主体身份对新闻记忆效果的影响。结果发现社会化线索和传播主体身份在影响受众收看新闻的记忆效果中均起到了显著作用，观看设置社会化线索的视频和观看由真人主播身份播报新闻的视频，受众的新闻记忆更好。
本文的第五章节，基于上述两项实验的总体结果，进行影响受众感知因素探究及应对策略探讨。共包含三部分内容，第一部分通过深度访谈法回应与补充实验研究的结论，分析受访者在观看实验材料中的具体感知与影响因素。第二部分通过实际案例选取，从“重音”与“停连”、“节奏”与“语速”、“语调”与“语气”、“音声性”与“规范性”等方面进行人工智能新闻主播与真人新闻主播的非言语符号差异性分析。第三部分提出人工智能新闻主播在结合播音学理论进行文本多维细分、 结合播音创作要素优化语义情感表达以及构建“物理场”、“心理场”、“审美场”的内部升级路径与人工智能新闻主播“形象构建与角色赋生”、“专业性输出与价值观赋能”以及“场域共频与圈层化联结”的外部升级策略探讨。 
本文的第六章节，在人工智能主播广泛应用的背景下提出对人的主体性与职业价值的再思考、对主流意识形态话语权构建中智能主播价值效能弱化问题以及可能产生的社会风险与伦理问题等关键议题进行理论层面的讨论，并提出实践层面的参考建议。
本文的第七章节为全文的总结和展望。通过上述研究，验证了媒介等同理论、恐怖谷效应对于人工智能主播这一新型传播主体的适用程度，以拓展理论的适用范围。同时，本实验制作了人工智能主播与真人主播播报新闻的视频材料，使受众的感知与评价基于真实的、直观的观看体验。弥补了先前实证研究中，由于被试者对人工智能主播的认知或理解不一致而导致结果偏差的可能性。本研究试图为受众对人工智能主播的接受机制与提升路径提供更多的理论视角与实证支撑。本论文的研究结论也有助于对人工智能主播未来的设计与应用、媒介智能机器生产的社会效应等问题提供更为广泛的理解与判断。

ABSTRACT：In recent years, artificial intelligence technology has driven journalism to turn toward intelligence, and the application of artificial intelligence technology in the field of journalism has shown the characteristics of a breakthrough from marginalization (terminal, equipment) to centralization (news production). As an important constituent element in the practical activities of news production, the production subject plays the key role of direct initiator, direct participant and direct presenter. Regardless of the media form, the announcer host always stands at the forefront of news production and information dissemination, directly facing the audience and becoming an important role in determining the communication effect. As the "spokesperson" of intelligent communication media representation symbols, the artificial intelligence anchor is the technical subject of the intelligent production platform "algorithm gate-keeping", and is the inevitable product of the deep integration development of new mainstream media under the support of artificial intelligence technology. Every innovation of human communication technology will bring changes in the form of media, changes in the relationship between people and information, changes in the way to obtain and feedback information, and changes in the vision and perception of all media as an extension of people.
In the first chapter of this paper, on the basis of the theoretical and social significance of the selected topic, the research question of this paper is proposed in a comprehensive analysis and summary of the existing related research. In this paper, we propose the research question of audience's emotional experience and cognitive effect on the news broadcast by AI anchors, including the "receiving-decoding-receiving" and "awareness-emotion-volition" processes. In the process of "receiving - decoding - receiving" and "awareness - emotion - will", does the emergence of AI anchors as communication subjects positively affect the results of "receiving"? What factors and indicators influence the audience's "receiving"? How does the change in the laws of natural and social attributes of the creative subject in communication affect the effect of communication?  
In the second chapter of this paper, firstly, the categories and characteristics of AI anchors are divided, and the research object of this paper is clearly the AI news anchors in China. Then, by sorting out the development, technical characteristics, application mode and development trend of AI anchors, we try to explore the social background and technical application of AI anchors in different periods, and select typical cases to explore the social motives behind the birth, development and withdrawal of AI anchors from the historical stage in different social backgrounds.
The third and fourth chapters of this paper explore the audience's perception, evaluation and memory effects on the news broadcast by AI anchors at this stage. Among them, the third chapter is centered on Experiment 1. Using the online experiment method, a 2 (anchor category: AI anchor vs. real anchor) × 2 (news type: livelihood news vs. current affairs news) × 2 (platform attribute: official media vs. self-published media) online experiment was conducted using independently recorded and produced videos of real anchors and AI anchors broadcasting news as experimental materials. The results show that anchor category, news type and platform attribute all significantly affect audience perception and evaluation. The news broadcasted by artificial intelligence anchors was lower than that of real anchors in terms of credibility, news immersion, audience liking, and overall quality evaluation. In addition, there was no significant difference between the overall quality ratings of livelihood news and current affairs news, but the trustworthiness and immersion of livelihood news were significantly higher than current affairs news and more liked by the subjects; there was no significant difference between the overall quality ratings of videos released by official media and self-published media, however, videos with official media logos were considered more trustworthy, more immersive and more liked by the subjects. 
The fourth chapter section of this paper is centered on Experiment 2. It investigates the memory effect of watching the news broadcast by an AI anchor and the difference in the memory effect of watching the news broadcast by a live anchor. The same controlled experimental method was used to record the same news broadcast by an AI anchor and a live anchor, based on the dual perspective of social cues and transmitter identity. In the social cue setting, the social cues were designed for the real anchor's physical system and vocal paralinguistic system, and the role of social cues was examined by comparing the video with and without the social cue setting; the effect of the identity of the communication subject on the news memory effect was also examined by questionnaire. The results showed that both social cues and the identity of the communicator played a significant role in influencing audiences' memory of watching news, and audiences remembered news better when they watched videos with social cues and videos with a live anchor.
In the fifth section of this paper, based on the overall results of the two experiments mentioned above, we explore the factors affecting audience perception and response strategies. The first part responds to and supplements the findings of the experimental study through in-depth interviews, and analyzes the specific perceptions and influencing factors of the respondents in watching the experimental materials. In the second part, the actual cases are selected, and the factors of "stress" and "stopping", "rhythm" and "speed of speech", "tone" and "speed of speech" are discussed, The second part analyzes the differences of non-verbal symbols between AI news anchors and real news anchors in terms of "accent" and "stop", "rhythm" and "speech speed", "tone" and "intonation", phonetics and normality. The third part proposes the internal upgrading paths of AI news anchors and artificial news anchors in terms of multi-dimensional subdivision of texts combined with broadcasting theory, optimization of semantic and emotional expressions combined with broadcasting creation elements, and construction of "physical field", "psychological field" and "aesthetic field". The internal upgrading path of "physical field", "psychological field" and "aesthetic field" and the external upgrading strategies of "image construction and role empowerment", "professional output and value empowerment" and "field co-frequency and circle connection" for AI news anchors are discussed. The external upgrading strategies of "image construction and role empowerment", "professional output and value empowerment" and "field co-frequency and circle connection" are discussed. 
The sixth chapter of this paper presents a theoretical discussion and practical suggestions on key issues such as rethinking human subjectivity and professional value, weakening the value of intelligent anchors in the construction of mainstream ideological discourse, and possible social risks and ethical issues in the context of the widespread use of artificial intelligence anchors.
The seventh section of this paper concludes and outlooks the whole paper. Through the above-mentioned research, the degree of applicability of media equivalence theory and the Valley of Terror effect to the new communication subject of AI anchors is verified in order to expand the scope of the theory. At the same time, this experiment produced video materials of AI anchors and real anchors broadcasting news, so that the audience's perception and evaluation are based on real and intuitive viewing experiences. The possibility of biased results due to inconsistencies in subjects' perceptions or understanding of AI anchors in previous empirical studies is compensated. This study attempts to provide more theoretical perspectives and empirical support for the acceptance mechanism and enhancement path of audiences towards AI anchors. The findings of this thesis also help to provide a broader understanding and judgment on the future design and application of AI anchors and the social effects of media intelligence machine production.