現代英語語言：

Author: Sun Yixiang, Yu Yuanlai
Abstract: The development and application of artificial intelligence technology has greatly changed the information dissemination and public opinion ecology, and also spawned the birth of “intelligent public opinion warfare”. Looking forward to the continued development of national discourse power and the technical competition behind the public opinion game of major powers, this article proposes the concept of “intelligent public opinion warfare” and interprets and analyzes its development prospects, application strategies and capacity building, aiming to explore the new frontiers, new games and new tactics of public opinion warfare in the era of intelligence, and provide theoretical reference for us to win the future public opinion battlefield.

Keywords: “intelligent public opinion warfare”; development prospects; application strategies; capacity building

“Intelligent public opinion warfare” is a new concept proposed based on the evolution of intelligent warfare forms and the development prospects of public opinion warfare. It is a new type of public opinion offensive and defensive action that uses new technologies and new applications in the field of artificial intelligence to spread political issues, derive discourse ecology, promote ideology, and influence public opinion. Its essence is to improve the speed, accuracy, intensity and breadth of information dissemination through the deep integration of disruptive technologies such as intelligent robots, deep learning, and algorithm recommendations with 5G, social media, big data, and the Internet of Things, thereby enabling traditional public opinion warfare to break through the “technical bottleneck” in core links such as situational awareness, target locking, information push, and effect evaluation, changing the phenomenon of weak targeting, limited aggregation, and weak controllability in some previous operations. It greatly exerts the “soft kill” effectiveness of attacking the mind and affecting cognition in response to the key and difficult problems such as high dependence on human intelligence and energy. The in-depth analysis of “intelligent public opinion warfare” aims to explore the new frontiers, new games, and new tactics of the development of public opinion warfare, and provide theoretical references for our army to build a new type of public opinion warfare force and win the future public opinion battlefield.

I. The development prospect of “intelligent public opinion warfare”

The essence of artificial intelligence is the simulation of human thinking. Its greatest advantage is to simulate, extend, and expand human intelligence. [1] Public opinion warfare takes media information attack and confrontation as its basic form, aiming to influence and persuade the target opponent and change his mental cognition. Its main “weapon platform” is various media for disseminating information. With the increasing maturity of artificial intelligence technology, the traditional media’s collection, writing, editing and broadcasting technologies have gradually been replaced by robot writers, algorithm recommendations, automatic broadcasting, audio-visual fitting, immersive news, etc. The subversive transformation of “weapon platforms” has also profoundly affected the development of public opinion warfare, making traditional public opinion warfare rapidly move towards a computable, modelable and predictable intelligent era. Various signs show that current intelligent technologies such as social robots, deep fakes, and big data analysis have been fully utilized in many major international events, and their influence on public opinion is becoming increasingly powerful.

(I) Artificial intelligence leads new changes in communication methods

Although current intelligent robots cannot fundamentally replace people as the main body of information dissemination, the information production and dissemination model of “artificial templates + automated data filling” and intelligent recommendation algorithms have become a bridge for more accurate docking between media and audiences. In addition, the in-depth popularization of software and hardware such as smart phones, smart chips, smart writing, virtual anchors, and AI editing has led to the global media industry actively seeking how to perfectly combine professional advantages with the efficient capabilities of artificial intelligence. For example, the robot Heliograf used by The Washington Post can automatically generate stories based on real-time data sources to provide audiences with personalized news. The robot Blossomblot used by the New York Times can screen and disseminate potential hot articles by analyzing social media data. [2] There are also writing software such as Kuaibi Xiaoxin used by Xinhua News Agency, Dreamwriter of Tencent, and Xiaomingbot of Toutiao. It is precisely based on the strong drive of technologies such as artificial intelligence, mobile Internet, big data, and blockchain that today’s media ecology has begun to develop from all-media and integrated media to smart media. [3] News push methods represented by algorithms and news production methods represented by robot news have become new features and changes in the development of the global communication field.

(II) Intelligent media gives new efficiency to public opinion manipulation

As an online social information dissemination system that combines artificial intelligence and human intelligence, intelligent media can create topics, attract fans, increase traffic, and continuously increase the influence and control of public opinion by creating virtual opinion leaders, analyzing user preferences with algorithms, innovating content distribution rules, and automatically publishing information in batches. At the same time, it can also carry out continuous and in-depth ideological penetration and public opinion influence, and intervene in the basic judgment and general cognition of the public. For example, during the 2016 US presidential election, Facebook used algorithms to construct user portraits and design software programs to predict and interfere with voters’ voting intentions and behaviors, and conduct political manipulation. It is precisely because of the remarkable characteristics of these intelligent media news production, personalized news release, on-the-spot news experience, and sensory interactive feedback that the “post-truth”, “information cocoon”, “echo chamber effect”, “spiral of silence”, “group polarization” and other phenomena in the public opinion field have become more prominent under human manipulation, which also provides new ways to win for all parties fighting in the public opinion battlefield. At present, in the process of producing massive press releases, intelligent media has developed into an important force for spreading political and military issues and conducting public opinion manipulation. According to the report of the “Computational Propaganda Research Project” of the Oxford Internet Institute, as of 2019, more than 50 countries in the world have found the use of robot water army to manipulate public opinion.

(III) Modern warfare highlights the new prospect of controlling public opinion with intelligence

Modern warfare has expanded from the tangible battlefield of the traditional physical domain to the intangible battlefield of the information domain and cognitive domain. Artificial intelligence technology gives state institutions great potential to manipulate public opinion quickly, cheaply and on a large scale, which also leads to the continued intensification of threats and challenges brought by new forms of warfare such as hybrid warfare and multi-domain warfare. Using intelligent communication means to efficiently spread emotional cognition, values ​​and opinions that are beneficial to oneself has become an important “booster” for achieving political and military goals in modern warfare. In this regard, former US Secretary of Defense Mark Esper once said: “The country that takes the lead in using artificial intelligence will have a decisive advantage on the battlefield for a long time in the future, so we must do it first.” For a long time, the US military has used a large number of false identities (i.e. “vests”) of “online virtual identity management” software to participate in chats or post on major social networking sites such as Facebook and Twitter to create pro-American remarks. Due to the low error rate and fully automated production, these robot water armies quickly became an important tool for influencing public opinion. In 2014, the US government also invested 760 million yuan to hire advertising companies to implement intelligent public opinion warfare specifically to combat targets such as al-Qaeda and the Somali Youth Party. In recent years, from major public opinion crises in Venezuela, Iran and other countries to the Syrian War and the Israeli-Palestinian conflict, it has become the norm to use social robots, deep fakes, virtual propaganda and other intelligent communication technologies and means to influence public opinion, and the “intelligent public opinion war” has gradually moved from behind the scenes to the front stage, showing a strong momentum of controlling public opinion with intelligence and winning with public opinion.

2. Application strategy of “intelligent public opinion war”

The mechanism of “intelligent public opinion war” is: with the help of machine learning, deeply explore the audience’s emotions and prejudices, screen and lock the most susceptible audiences, and then use algorithms to produce massive persuasive content, which is then flooded, disseminated and forwarded in batches by online news aggregators, corporate customer service robots, etc., thereby destroying the original international discourse system and communication ecology, and achieving the purpose of manipulating public opinion. The subject, object, topic, content, means and other key elements of its public opinion war are highly dependent on the empowerment and efficiency of intelligent technology. Nowadays, major countries are actively seeking to seize the dominant discourse power and win the initiative in the war of public opinion through intelligent “technological gap”. The main strategies used are:

(I) Cultivating virtual communication subjects

The implementation subjects of traditional public opinion warfare include not only propaganda departments, media organizations, think tanks, but also press spokespersons, opinion leaders and ordinary netizens. Intelligent technology has expanded this subject to a large number of “human-machine mobs” such as writing robots, dialogue robots, virtual bloggers and water armies, and has shown a blowout trend worldwide. According to the US “Capsule” website, the “2016 Machine Traffic Report” released by the Internet security company IMPV stated that they analyzed the access records of 100,000 domain names worldwide and found that about 52% of the Internet traffic came from “automated programs”, that is, robots. The Pew Research Center in the United States also found in a 2018 survey that 66% of the 1.2 million Twitter posts linking to 2,315 popular websites collected from various social media platforms were shared by robots. There is also the “troll army” created by Russia, which deploys a large number of robot virtual accounts. Its Internet Research Bureau can push out 25,000 tweets within 24 hours. These emerging forces have led to the continued intensification of the “asymmetric” situation of public opinion wars between countries. Studies have shown that people’s ideas are easily influenced by the information they see online, and the comments made by robots on social networks can often have an important impact on their ideas.

(II) Deeply dig and lock in the target audience

The target audience is the target of public opinion wars, and quickly identifying the social characteristics, position tendencies and lifestyles of the target audience will help improve the ability to target attacks based on the cultural psychology of the audience, and actively shape and change the group cognition of the target object. The “cloud brain” built with intelligent algorithms can deeply analyze the user’s reception needs and interest trends, summarize behavioral models, depict user portraits, discover internal laws, and determine the most accurate content, best time, and best path for information push, thereby improving the accuracy and effectiveness of precise segmentation and differentiated communication. At the same time, by grasping the differentiated needs of the target audience through intelligent technology, it is also possible to realize the intelligent construction of different communities and circles from content production to distribution, so as to match the value orientation of different audience groups and meet their personalized needs. Such communication advantages help to better integrate diverse and scattered viewpoints and quickly gather the leading opinions of public opinion. At present, in addition to intelligent social media, smart cars, smart shopping platforms, etc. have become important platforms for user information collection. According to the data provided by these platforms, it is easy to draw key social structures, human maps, as well as key influencers and opinion leaders, which also makes the target of public opinion war more accurate and more efficient.

(III) Automatic aggregation of hot topics

Setting topics is an important link in strengthening public opinion control. Intelligent technology strengthens the mining, aggregation and sharing functions of hot topics in massive information, so that the human factors in the fermentation process of public opinion continue to rise, and the breeding and development of public opinion are more likely to develop in the direction expected by the manipulator. On the one hand, big data mining news clues can quickly refine, edit and publish topics that attract public attention and trigger public discussion. For example, the US NewsWhip uses big data mining technology to scan major global social media such as Facebook, Twitter, and Instagram every two minutes to capture what events are the hottest and what content is more popular. It then analyzes the development trends of these contents and converts them into actionable suggestions to help media professionals plan topics based on clues. [4] On the other hand, intelligent communication media can enable different information products such as audio, video, images, and text to be freely matched and shared in the form of data information flow on wearable devices, cloud terminals, websites, and related social media, which further intensifies the aggregation and influence of hot public opinion.

(IV) Using algorithms to generate content

As the “ammunition” of public opinion warfare, information content has a distinct direction, pertinence, and strong topicality. Its planning, production, and packaging are mainly based on political and military objectives, rather than pure news value or commercial value. As the node between data information and artificial intelligence, algorithms play a key role in constructing traffic entrances and capturing user stickiness. For example, the algorithm distribution mechanism used by most current news apps and social media platforms can capture and analyze natural information such as the user’s age, gender, occupation, and cultural level, as well as behavioral information such as contact duration, contact frequency, and contact motivation during reading, through machine learning, to achieve accurate user portraits. At the same time, various “news materials” are collaged according to instructions and pushed to these pre-targeted target audiences in a targeted manner. In the process of bringing high-quality experience and meeting individual needs, it will also aggravate the “information cocoon effect”, trapping these audiences in the “cocoon”. Under the influence of a large amount of homogeneous content, their thoughts and cognition will change subtly. For this reason, the article “Artificial Intelligence and National Security” issued by the US Congress in November 2020 mentioned that it is necessary to create a complete “digital life model” to match and combine the target audience’s shopping records, credit reports, professional resumes, subscriptions, etc., to form a comprehensive behavioral profile of them, so as to better serve the influence action.

(V) Implement deep fake deception interference

The battlefield of public opinion has always been a combination of reality and fiction, and the flexible use of various “fog” and “traps” is a common means of winning for all parties in the game. In recent years, the hot deep fake technology, as a new means of information tampering and replacement based on deep learning, can generate highly realistic and difficult to identify information products through voice simulation, face synthesis, video generation and other methods, which also makes it a “throat-locking” trick in the battlefield of public opinion. For example, in 2018, the American online news media “Buzz Feed” made a fake video of Obama insulting Trump by grafting the voice and mouth shape of comedian Jordan Peele. Once it was launched on YouTube, it received 4.8 million hits. This video used artificial intelligence face-changing technology. In addition to face-changing videos, deep fakes are also often used for portrait synthesis. Compared with traditional character images and video synthesis, it has significant characteristics such as low cost, low operation difficulty, and short time required. This also makes the use of deep fake technology to create negative videos of politicians, false orders from military commanders, violent terrorist information and other “new rumors” emerge in an endless stream in today’s public opinion field. For this purpose, the US military has also specially carried out the research and development of projects such as “Trojan Horse in Artificial Intelligence” and “Reliable Guarantee of Artificial Intelligence for Deception”.

3. Capacity Building of “Intelligent Public Opinion Warfare”

“Intelligent public opinion warfare” is not only a contest of technology, but also a contest of capabilities. Although “weak artificial intelligence” technology is still widely used in the field of public opinion warfare, with the rapid development of science and technology, the capabilities of artificial intelligence will continue to increase, and public opinion warfare will also usher in a new era of intelligence. Looking forward to the major strategic opportunities for the development of artificial intelligence in my country, the continued development of national discourse power and the practical needs of responding to the attacks of the United States and the West on our public opinion warfare, it is necessary to strengthen multidisciplinary integration research, cross-exploration from multiple perspectives, and actively promote the transformation, upgrading and rapid development of “intelligent public opinion warfare” in technology, theory, practice, etc.

(I) Increase the technical research and development of “intelligent public opinion warfare”

“Intelligent public opinion warfare” must not only compete with people’s hearts, but also compete with machines. To this end, it is necessary to take policy as the guide and science and technology as the driving force to step up the research and development of intelligent technology in the field of international communication and public opinion attack and defense. First, we will vigorously solve the outstanding problems in media applications, such as framework bias, machine cognitive bias, and machine communication ambiguity, and fully activate the deep integration of artificial intelligence with 5G, social media, big data, the Internet of Things, virtual reality and other technologies in information production, presentation, and dissemination.
The first is to create a new driving force in the field of “all-media journalists + algorithm engineers” and “killer” weapons for public opinion warfare to meet the actual needs of public opinion games in the intelligent era. The second is to strengthen the in-depth application of artificial intelligence technology in media interviews, writing, editing, and even robot hosting and broadcasting, deeply integrate the content production advantages accumulated by traditional media for a long time with intelligent technology, and comprehensively use text, audio and video, animation, live broadcast, data illustration and other forms to innovate the attractiveness and communication power of public opinion propaganda content; at the same time, distinguish different target objects to formulate specific communication strategies, improve the information product structure, enrich product formats, and accurately and segmentally influence the ideological cognition of each specific target group.

(II) Deepen the theoretical research on “intelligent public opinion warfare”

“Intelligent public opinion warfare” must not only solve the relationship between “construction” and “use”, but also grasp the laws of “root” and “end”, and focus on strengthening the exploration and dissection of the connotation mechanism, and effectively provide scientific theoretical guidance for promoting “intelligent public opinion warfare”. First, from the perspective of “public opinion + technology”, study how public opinion warfare can empower and increase efficiency. We must not only think about the problem from the perspective of competing with the enemy for the cognitive domain and the right to speak in public opinion, but also grasp the development trend of artificial intelligence communication technology, integrate advanced technical means into key links such as public opinion war situation perception, target selection, information dissemination, and effect evaluation, and deeply explore the winning mechanism, operational command, and capacity building of “intelligent public opinion war”. Second, from the perspective of “tactics + algorithm”, study how public opinion war can be fought intelligently. We must focus on the characteristics of the future public opinion battlefield such as “unmanned, invisible”, “brain-like, brain-controlled”, “actuarial, cloud computing”, and strengthen the application of advanced concepts such as intelligent game, cloud brain victory, visual expression, and asymmetric checks and balances; at the same time, summarize the superb strategies of struggles in the fields of military, politics, and public opinion in ancient and modern times, China and foreign countries, especially the experience and losses of using intelligent technology to carry out public opinion wars in recent years around the world, inherit in criticism, develop in innovation, and continuously improve the strategies and tactics that adapt to the development needs of “intelligent public opinion war” and have Chinese characteristics.

(III) Promote the practical application of “intelligent public opinion war”

The battlefield of public opinion is also wartime in peacetime, and it is still the same in the intelligent era. We must focus on daily life and the ubiquitous public opinion game to increase the intensity of tempering. First, optimize the overall layout and practical capabilities of intelligent public opinion dissemination, and gradually eliminate the “time difference” and “technical gap” with the capacity building and tactics of the United States and the West. Especially in the current situation where the United States and Western countries use the new crown pneumonia epidemic, the South China Sea, Taiwan, Hong Kong, Xinjiang and other issues to intensively use intelligent means to smear and attack me, we must constantly sum up experience, find ways and improve our capabilities in actual combat. Second, accurately grasp the technical means and strategic methods of hostile forces at home and abroad to penetrate and break through my public opinion space, and promptly cut off the internal and external coordination and transmission channels of anti-China forces, and prevent the spread of politically harmful information and the cognitive influence on me. Third, we will increase the comprehensive use of national defense mobilization, administrative law, public opinion guidance and other means, integrate the propaganda, public security, diplomacy, intelligence, military and other departments, and cooperate with the media, think tanks, international communication public relations companies, opinion leaders, etc., and actively reserve combat power for the “intelligent public opinion war” by seeking the first mover advantage in all directions.

Notes:

[1] Chen Changfeng, Huo Jie. People-oriented: The application of artificial intelligence technology in the field of news communication [J]. News and Writing, 2018(8):54.

[2] Wan Ke. The application practice and enlightenment of artificial intelligence in American and British news media [J]. China Media Technology, 2017(7):19-20.

[3] Li Peng, Zhi Media. The new stage of media integration and transformation [N]. People’s Daily Online, 2019-04-19.

[4] Xinhua News Agency. “Media Transformation and Development in the Era of Artificial Intelligence” Research Group. The specific application of artificial intelligence in the entire chain of news communication [J]. China Journalist, 2020(2):19.

(Author’s unit: Political College of National Defense University. This article is a phased result of the research project of the National Social Science Fund.)

現代國語：

作者：■孫亦祥 餘遠來
摘 要：人工智慧技術的發展運用，大大改變了資訊傳播、輿論生態，也催生了「智慧化輿論戰」的誕生。前瞻國家話語權的續存發展和大國輿論博弈背後的技術較量，本文提出了「智能化輿論戰」概念，並圍繞其發展圖景、運用策略和能力構建進行闡釋分析，旨在探究智能化時代輿論戰發展的新疆域、新賽局和新戰法，為我制勝未來輿論戰場提供理論參考。

關鍵字：「智能化輿論戰」；發展圖景；運用策略；能力建構

「智能化輿論戰」是基於智慧化戰爭形態演變和輿論戰發展前景而提出的新概念，是利用人工智慧領域新技術新運用擴散政治議題、衍生論述生態、推廣意識形態、影響公眾意見的一種全新輿論攻防行動。其實質是透過智慧機器人、深度學習、演算法推薦等顛覆性技術運用與5G、社群媒體、大數據、物聯網等深度融合來提升資訊傳播的速度、精度、強度和廣度，進而使傳統輿論戰在態勢感知、目標鎖定、資訊推送和效果評估等核心環節得以突破“技術瓶頸”，改變以往有的作戰中靶向性不強、聚合力有限、可控性偏弱等現象，對於人的智力、精力依賴程度高等重難點問題，大幅發揮攻心奪志、影響認知的「軟殺傷」效能。對於「智能化輿論戰」的深入分析，旨在探索輿論戰發展的新疆域、新博弈和新戰法，為我軍打造新質輿論戰力量，制勝未來輿論戰場提供理論參考。

一、「智能化輿論戰」的發展圖景

人工智慧的本質是人類思維的模擬，它的最大優勢是模擬、延伸和擴展人的智慧。 [1]而輿論戰以媒介訊息攻擊與對抗為基本形式，旨在影響、說服目標對手，改變其精神認知，它的主要「武器平台」就是傳播訊息的各類傳媒。伴隨人工智慧技術的日益成熟，傳統媒體的採、寫、編、播技術開始逐步被機器人寫手、演算法推薦、自動播報、聲像擬合、沉浸式新聞等替代，「武器平台」的顛覆性變換也深刻影響輿論戰的發展形態，使傳統輿論戰開始朝著可計算、可建模、可預測的智慧化時代快速邁進。各種跡象表明，當前社交機器人、深度偽造、大數據分析等智慧化技術已在許多國際重大事件中充分運用，其輿論影響日益強大。

（一）人工智慧引領傳播方式新變化

儘管目前智慧機器人還不能從根本上替代人成為傳播訊息的主體，但是「人工模板+自動化資料填充」的訊息生產傳播模式，以及智慧推薦演算法已成為媒體與受眾間更精準對接的橋樑紐帶。加之智慧型手機、智慧晶片、智慧寫稿、虛擬主播、AI剪輯等軟硬體的深入普及，導致全球傳媒產業都在積極尋求如何將專業優勢與人工智慧的高效能力進行完美結合。如《華盛頓郵報》所使用的機器人Heliograf，可根據即時資料來源自動產生故事，並為受眾提供個人客製化新聞。 《紐約時報》所使用的機器人Blossomblot，可透過分析社群媒體資料來篩選傳播潛在的熱文等。 [2]還有國內新華社使用的快筆小新、騰訊的Dreamwriter、今日頭條的Xiaomingbot等寫作軟體。正是基於人工智慧、行動互聯網、大數據、區塊鏈等技術的大力驅動，現今的媒體生態已由全媒體、融媒體開始向智媒體發展。 [3]以演算法為代表的新聞推送方式和以機器人新聞為代表的新聞生產方式已成為全球傳播領域發展的新特色新變化。

（二）智能傳媒賦予輿論操控新效能

智慧媒體作為一種人工智慧與人類智慧協同的線上社會資訊傳播系統，透過創建虛擬意見領袖、演算法分析使用者偏好、革新內容分發規則、自動大量發布資訊等手段，在製造話題、吸引粉絲、提升流量，不斷加大輿論影響力控制力的同時，還能進行持續深入的思想滲透和輿論影響，幹​​預社會公眾的基本判斷和普遍認知。如2016年美國大選期間，Facebook透過演算法建構用戶畫像並設計軟體程序，從而預測和乾涉選民的投票意向和行為，進行政治操控等。正是由於這些智慧媒體新聞生產的機器化、新聞發布的個人化、新聞體驗的臨場化、互動回饋的感測化等顯著特點，使得輿論場的「後真相」「資訊繭房」迴聲室效應」「沉默的螺旋」「群體極化」等現像在人為操控下變得更為突出，這也為廝殺輿論戰場的各方提供了新的製勝途徑。目前智慧傳媒在生產大量新聞稿的過程中，已發展成為傳播政治軍事類議題，並進行輿論操控的重要力量。根據牛津大學網路研究所「計算宣傳研究計畫」報告稱，截至2019年，全球已有50多個國家發現運用機器人水軍操縱輿論的情況。

（三）現代戰爭凸顯以智制輿新前景

現代戰爭已從傳統物理域有形戰場拓展到資訊域、認知域的無形戰場。而人工智慧技術賦予國家機構迅速、廉價、大規模操縱輿論的巨大潛能，也導致混合戰、多域戰等新型戰爭形態帶來的威脅挑戰持續加劇。借助智慧化傳播手段高效傳播有利於己的情感認知、價值觀念和意見傾向，已成為現代戰爭政治軍事目標實現的重要「助推器」。對此美國前國防部長馬克·埃斯珀曾說：「率先使用人工智慧的國家，將在未來很長時期的戰場上擁有決定性優勢，因此我們必須首先做到。」長期以來，美軍利用「在線虛擬身分管理」軟體的大量虛假身分（即「馬甲」），在Facebook、Twitter等各大社群網站上參與聊天或發文以製造親美言論。由於出錯率低、生產完全自動化，這些機器人水軍很快就轉變為影響輿論的重要工具。 2014年美國政府也出資7.6億元聘請廣告公司，專門為打擊蓋達組織、索馬利亞青年黨等目標而實施智能化輿論戰。近年來，從委內瑞拉、伊朗等多個國家遭遇的重大輿論危機事件到敘利亞戰爭、巴以衝突來看，利用社交機器人、深度偽造、虛擬宣傳等智能化傳播技術與手段來影響公眾輿論已成常態，而「智能化輿論戰」也逐漸從幕後走向前台，並呈現出以智駕馭輿、以輿制勝的強勁勢頭。

二、「智能化輿論戰」的運用策略

「智慧化輿論戰」的作用機制是：藉助機器學習深度挖掘受眾情感、偏見並篩選、鎖定最易受影響的受眾，然後利用演算法生產大量說服性內容，並由網路新聞聚合器、企業客服機器人等進行批量灌水、散佈和轉發，從而破壞原有的國際話語體系和傳播生態，達到操控輿論的目的。其輿論戰的實施主體、客體，以及議題、內容、手段等關鍵要素都高度依賴智慧化技術的賦能增效。在當今，各主要國家都在積極謀求透過智慧化「技術差」來搶佔優勢話語權，贏得輿論戰主動權，主要運用策略有：

（一）培塑虛擬傳播主體

傳統輿論戰的實施主體既有宣傳部門、媒體機構、智庫，也有新聞發言人、意見領袖和一般網友等。而智慧化技術讓這個主體拓展到了寫稿機器人、對話機器人、虛擬部落客和水軍等大批“人機烏合之眾”，並在全球範圍內呈現井噴之勢。據美國“膠囊裡”網站報道，互聯網安全公司IMPV發布的《2016年機器流量報告》稱，他們通過對全球10萬個域名網站訪問記錄分析發現，其中約52％的互聯網流量來自“自動化程序” ，即機器人。美國皮尤研究中心2018年調查也發現，它們從各類社群媒體平台收集到的連結至2315個流行網站的120萬則Twitter貼文中，有66%為機器人所分享。還有俄羅斯打造的“巨魔大軍”，部署有海量機器人虛擬帳號，其互聯網研究局在24小時內就能推出2.5萬條推文。這些異軍突起的新興力量，導致各國間進行輿論戰的「非對稱」態勢持續加劇。有研究顯示，民眾的想法很容易受到他們在網路上所看到的資訊的影響，而社群網路上的機器人發表的言論往往能夠對他們的想法施加重要的影響。

（二）深挖鎖定目標受眾

目標受眾是輿論戰的靶標，而快速識別目標受眾的社會特性、立場傾向和生活方式等，將有助於提高基於受眾文化心理的靶向打擊能力，積極塑造和改變目標對象的群體認知。以智慧演算法建構的「雲端大腦」能夠深刻解析使用者接收需求和興趣動向，總結行為模型、描繪使用者畫像、發現內在規律，並確定訊息推送的最準內容、最佳時機、最優路徑，提高精準分眾、差異傳播的準確率和致效強度。同時，透過智慧技術掌握目標受眾的差異化需求，也能夠對不同社群、圈群實現從內容生產到分發的智慧化構建，以契合不同受眾群體的價值傾向，滿足他們的個人需求。這樣的傳播優勢有助於更好地整合多元、分散觀點，快速匯聚輿論主導意見。目前除智慧社交媒體外，智慧汽車、智慧購物平台等都已成為用戶資訊收集的重要平台。根據這些平台提供的數據很容易繪製出關鍵性社會結構、人文地圖，以及關鍵影響者和輿論意見領袖，這也使得輿論戰的標靶變得更精準，致效率變得更高。

（三）自動聚合熱點議題

設定議題是強化輿論控制的重要環節，智能化技術強化了海量資訊中的熱點議題挖掘、聚合和分享功能，從而使輿論發酵過程中的人為因素持續攀升，輿論的滋生發展更容易朝操控者所期待的方向發展。一方面，大數據挖掘新聞線索，能夠快速提煉、編輯和發布吸引公眾眼球，引發眾議的議題。如美國NewsWhip（新聞鞭）透過大數據挖掘技術，可以每兩分鐘掃描一次Facebook、Twitter、Instagram等全球主要社群媒體，來捕捉什麼事件是當下最熱門的，怎樣的內容更受歡迎，然後分析這些內容的發展趨勢並轉化為可操作的建議，幫助媒體人根據線索進行選題策劃。 [4]另一方面，智慧化傳播介質能夠讓音訊視訊、影像、文字等不同資訊產品得以以資料資訊流的形態在穿戴式裝置、雲端終端、網站及相關社交媒介中進行自由配對與分享，這也進一步加劇了熱點輿論的聚合力影響力。

（四）利用演算法生成內容

資訊內容作為輿論戰的“彈藥”，具有鮮明的指向性、針對性和強烈的話題度，它的策劃、製作和包裝主要基於政治、軍事目標的考量，而非單純的新聞價值或商業價值。而演算法作為資料資訊與人工智慧的節點，發揮建構流量入口、捕捉用戶黏性的關鍵作用。如當前大多數新聞App和社群媒體平台使用的演算法分發機制，其透過機器學習，抓取和分析用戶的年齡、性別、職業、文化水平等自然訊息，以及閱讀時的接觸時長、接觸頻率、接觸動機等行為訊息，可以實現對使用者的精準畫像。同時，根據指令拼貼各種“新聞素材”，有針對性地推送給這些預先鎖定的目標受眾。在帶來優質體驗、滿足個性需求的過程中，也會加劇“資訊繭房效應”，使這些受眾困於“繭房”之中，在大量同質化內容的灌輸影響下，思想認知發生潛移默化的改變。正因如此，美國國會於2020年11月頒發的《人工智慧與國家安全》一文中提及，要透過創造完整的“數位生活模型”，將目標受眾的購物記錄、信用報告、職業簡歷、訂閱等匹配組合，對其形成綜合性的行為簡介，從而更好地服務於影響力行動。

（五）實施深偽欺騙幹擾

輿論戰場向來虛實結合，各種「迷霧」「陷阱」的靈活運用是博弈各方慣用的製勝手段。近年來，大熱的深度偽造技術，作為基於深度學習的一種新型信息篡改與替換手段，透過語音模擬、人臉合成、視頻生成等方法可以生成高逼真且難以甄別的信息產品，這也使其成為了輿論戰場的「鎖喉」招數。如2018年美國網路新聞媒體「嗡嗡餵」透過嫁接喜劇演員喬丹皮爾的聲音和嘴型，製作了一段奧巴馬辱罵川普的假視頻，YouTube一經上線就獲得了480萬次的點擊量，這段影片運用的就是人工智慧換臉技術。除換臉影片外，深度偽造也常被用於人像合成，與傳統的人物影像、影片合成相比，它具有使用成本低、操作難度小、所需時間短等顯著特點，這也使得利用深度偽造技術製造政治人物負面影片、軍隊指揮官虛假命令、暴力恐怖訊息等「新型謠言」在當今輿論場層出不窮。美軍為此也特別進行了「人工智慧中的特洛伊木馬」「面向欺騙的人工智慧可靠保障」等項目的研發。

三、「智能化輿論戰」的能力建構

「智能化輿論戰」既是技術的較量，也是能力的比拼。儘管目前輿論戰領域普遍應用的還是「弱人工智慧」技術，但是隨著科技的快速發展，人工智慧的能力將不斷增強，輿論戰也必將迎來全新的智慧化時代。前瞻我國人工智慧發展的重大戰略機遇，國家話語權的續存發展和應對美西方對我輿論戰攻擊的現實需要，必須加強多學科的融合研究，多視角的交叉探索，積極推動「智能化輿論戰」在技術、理論、實踐等方面的轉型升級、快速發展。

（一）加大「智能化輿論戰」的技術研發

「智能化輿論戰」不僅要與人心博弈，更要與機器較量，為此必須以政策為導向、科技為牽引，加緊推動智能化技術在國際傳播和輿論攻防領域的研究攻關。一是大力破解媒體應用方面依然存在的框架偏差、機器認知偏差、機器交流歧義等多方面突出問題，全面激活人工智能與5G、社交媒體、大數據、物聯網、虛擬實境等技術的深度融合和在資訊生產、呈現、傳播等面向的全新驅動力，系統打造「全媒體記者+演算法工程師」與輿論戰「殺手鐧」武器，以因應智慧化時代輿論博弈的現實需求。二是加強人工智慧技術在媒體採訪、寫稿、編輯，甚至機器人主持播報諸多領域的深化運用，將傳統媒體長期累積的內容製作優勢與智慧科技深度融合，綜合運用文字、影音、動漫、直播、數據圖解等形式，創新輿論宣傳內容的吸引力傳播力；同時區分不同目標對象制定特定傳播策略，完善資訊產品結構、豐富產品業態，精準化、分眾化影響各特定目標群的思想認知。

（二）深化「智慧化輿論戰」的理論攻關

「智能化輿論戰」既要解決好「建」與「用」的關係，也要把握好「本」與「末」的規律，要注重加強內涵機理的探索解剖，切實為推進「智能化輿論戰”提供科學的理論指導。一是從「輿論+技術」的角度出發，研究輿論戰如何賦能增效。既要站在與敵爭奪認知域、輿論話語權的高度思考問題，又要掌握人工智慧傳播技術的發展趨勢，將先進技術手段融入輿論戰態勢感知、目標選擇、資訊傳播、效果評估等重點環節，深入探索「智慧化輿論戰」的致勝機制、運籌指揮與能力建構等。二是從「戰法+演算法」的角度出發，研究輿論戰如何智慧作戰。要聚焦未來輿論戰場「無人、無形」「類腦、控腦」「精算、雲算」等特點，強化智能博弈、雲腦制勝、可視化表達、非對稱制衡等先進理念的運用；同時總結古今中外軍事、政治、輿論諸領域鬥爭的高超謀略，尤其是全球範圍內近年運用智能技術開展輿論戰的經驗得失，在批判中繼承，在創新中發展，不斷完善適應“智能化輿論戰”發展需要且具有中國特色的謀略戰法。

（三）推動「智能化輿論戰」的實戰運用

輿論戰場的平時也是戰時，智慧化時代依然如此，必須立足日常、著眼無時無所不在的輿論博弈加大錘煉砥礪力度。一是優化智慧化輿論傳播的整體佈局和實戰能力，逐步消除與美西方能力建設和戰法運用的「時間差」「技術差」。特別在當前美西方國家利用新冠肺炎疫情、南海、台灣、香港、新疆等問題，密集使用智能手段對我進行輿論抹黑攻擊的情況下，要在實戰應對中不斷總結經驗、尋找辦法、提升能力。二是精準掌握境內外敵對勢力向我輿論空間進行滲透突破的技術手段與策略手法，及時斬斷反華勢力內外策應傳導渠道，阻遏政治有害資訊傳播及對我實施認知影響。第三是加大國防動員、行政法、輿論引導等手段的綜合運用，宣傳、公安、外交、情報、軍事等部門的一體聯動，以及媒體、智庫、國際傳播公關公司、意見領袖等的配合與協作，透過全方位謀求先發之勢，為「智能化輿論戰」積極儲備戰力。

註 釋：

[1]陳昌鳳、霍婕.以人為本：人工智慧技術在新聞傳播領域的應用[J].新聞與寫作，2018(8):54.

[2]萬可.美英新聞媒體人工智慧應用實務及啟示[J].中國傳媒科技，2017(7)：19-20.

[3]李鵬、智媒體.媒體融合轉型新階段[N].人民網，2019-04-19.

[4]新華社.「人工智慧時代媒體變革與發展」主題組.人工智慧在新聞傳播全鏈條中的具體應用[J].中國記者，2020（2）：19.

（作者單位：國防大學政治學院。本文為國家社科基金計畫研究階段性成果。）

中國軍事原文來源：http://www.81.cn/rmjz_203219/jsjz/2022nd1q_227501/yldzgzyj_227505/10145288.html