Staff at a short drama production company in Jiaxing Economic and Technological Development Zone, Zhejiang Province, are engaged in the full workflow of script polishing, AI post-production, and overseas translation and distribution. Photo: VCG

AI short videos have rapidly gained public visibility thanks to their ease of production, striking realism, and capacity to reach wide audiences. Historical figures appear to speak, and scientific demonstrations look almost real. Short videos produced with the help of generative AI are changing the way people see the world, understand reality, and imagine the future. Their significance extends beyond a new category of media content, warranting scrutiny through communication studies, media theory, and social cognition.

On the surface, AI short videos represent a new, technologically enabled form of video production. At a deeper level, however, AI’s role has shifted from serving as a tool to generating simulacra—images that appear real but have no original counterpart in reality. Rather than depicting an existing object, algorithms statistically reorganize and re-encode vast amounts of visual cultural data to create these images. In other words, AI short videos do more than passively record reality: they actively generate, piece together, and substitute for it. It is in this sense that they constitute an “intelligent spectacle” worthy of serious discussion.

Conceptual shift

Research on AI short videos has largely developed through two stages. The first adopts a view of AI as a tool that assists with short video production, including script generation, intelligent editing, subtitle optimization, and content recommendation. This perspective emphasizes greater efficiency and smoother workflows, treating AI as a means of helping creators produce videos faster and more accurately. The second adopts a generative or ontological perspective. As text-to-video models and multimodal generation technologies have matured, AI has moved beyond assisting creators to generating video content directly. The fundamental shift is that this content derives from models’ probabilistic learning and reconstruction of training data rather than footage of the real world. Its apparent realism thus rests on algorithms’ ability to imitate visual experience rather than on a direct connection to the reality depicted. This shift turns AI short videos into a system of image production capable of generating meaning independently, extending their role beyond that of a technical tool.

Understanding this shift requires looking beyond technological advances in the short video industry to the broader questions raised by media philosophy and the politics of communication. AI short videos challenge both established methods of content production and the basic conceptual distinctions through which people understand reality, images, and representation.

How AI short videos reshape collective cognition

Traditional media research often uses the metaphor of a “social mirror” to describe the relationship between media and reality—media reflects society as a mirror would. Yet this metaphor cannot adequately explain how AI short videos construct layers of experience. These videos actively produce social meaning by selecting, combining, reorganizing, and amplifying social experiences, or even fabricating them entirely. “Social landscape” offers a more suitable way to understand their influence. The term describes the interconnected understandings of reality, values, and possible futures that a community develops through media, culture, and daily practices. Its emphasis falls on the ongoing construction of a shared picture of society rather than a static reflection of it.

AI short videos generate a version of reality for viewers to watch, which platform algorithms then deliver to individual users through personalized recommendations. Instead of encountering a shared public world of images, users inhabit separate “information cocoons” tailored by algorithms. Over time, their exposure to social reality depends increasingly on what platforms distribute rather than on direct experience. Social imagery is consequently fragmented and reorganized, becoming highly personalized, emotionally charged, and even disjointed.

Beyond changing the speed and form of information dissemination, AI short videos shape the deeper processes through which the public understands the world, distinguishes truth from falsehood, and reaches a consensus on values.

Generation, distribution, and immersion

The way AI short videos operate can be understood theoretically as an “intelligent spectacle,” extending French theorist Guy Debord’s theory of the “spectacle” and French philosopher Jean Baudrillard’s theory of “simulacra and simulation” into the digital age. In this account, the spectacle consists of social relationships mediated by images rather than simply an accumulation of images. In a postmodern consumer society, simulacra obscure and replace the reality they once represented, ultimately producing hyperreality. AI short videos bring these two theoretical perspectives together: algorithms automatically generate video content, while platform recommendation systems deliver it to precisely targeted audiences. Together, these processes create a continuously self-reinforcing cycle of spectacle.

Three features distinguish the spectacle of AI short videos from traditional spectacles. First, production is automated, with models and prompts generating content previously planned and filmed primarily by humans. Second, distribution is personalized, targeting content according to user profiles rather than addressing a single mass audience. Third, immersion reinforces engagement as repeated viewing, liking, sharing, and commenting draw users further in, fostering emotional dependence and habitual patterns of use.

The intelligent spectacle of AI short videos thus constitutes an evolving visual order, continuously updated and generated on demand, rather than a static collection of images. Its influence extends beyond directing attention and viewing habits to shaping perception itself and the judgments people make about reality.

Cognitive and ethical risks

One of the most troubling effects of AI short videos is a growing uncertainty about the boundary between reality and fiction. As viewers repeatedly encounter generated content, their ability to distinguish the two becomes less stable. They may recognize that a video is fictional yet still become emotionally invested in it, gradually internalizing what it depicts as a plausible model of reality. Both psychological responses to media and the ways platforms distribute content contribute to this blurring. Highly realistic images often elicit emotions similar to those prompted by actual experiences, and greater visual realism makes them more persuasive. The mere exposure effect helps explain how repeated recommendations of similar content can lower viewers’ guard, making it seem familiar, reasonable, and even real. Meanwhile, recommendation algorithms continually reinforce particular styles of content based on user preferences. The resulting cognitive closed loop makes it increasingly difficult for individuals to encounter information that challenges their existing judgments.

At the societal level, this blurring of reality and fiction poses several risks. First, it can erode trust—deepfakes and AI recreations of historical figures may disrupt public memory and undermine social trust. Second, it can standardize aesthetic preferences and models of living, as a proliferation of AI content depicting perfect lives breeds dissatisfaction and anxiety about real life. Third, it can weaken individual agency. With continued exposure to algorithmically selected content, individuals increasingly rely on platforms to determine what they watch and which values guide them, potentially diminishing their capacity for independent judgment and creativity.

From technological regulation to cognitive repair

Addressing the new challenges posed by AI short videos requires governance that moves beyond general warnings to a systematic framework for action at multiple levels.

First, content labeling and source tracing should be strengthened. AI-generated videos should be clearly identified, with traceable digital watermarks and records of the generation process to help viewers distinguish authentic content from synthetic material. Second, platforms should be held more accountable for their algorithms. Their responsibilities extend beyond promoting content to addressing the biases, polarization, and immersive effects of their recommendation systems. Finally, media literacy education should be improved to help the public recognize AI-generated content, understand the logic of algorithms, and view content critically.

Governance must also address the long-term influence of AI short videos on collective cognition alongside their technical risks. Establishing content authenticity is only part of the task. Equally necessary is restoring the public’s ability to judge reality, the basis for agreement on shared facts, and the foundations of trust among individuals.

Ultimately, the rise of AI short videos marks a shift in image production from recording reality to generating it, and from facilitating communication to organizing cognition. The production of simulacra, their algorithmic distribution, and the personalized immersion they enable gradually give rise to a new intelligent spectacle. Within it, reality ceases to be the sole point of reference, while fiction acquires influence beyond its imagined domain. Reality and fiction are continually blended and reconfigured through interactions among platforms, algorithms, and users, becoming embedded in the underlying structure of social imagery.

AI short video technologies have moved from tools for representing reality to generators of simulacra. The corresponding shift from “social mirror” to “social landscape” underscores their active role in constructing society as well as reflecting it.

The significance of AI short videos therefore extends beyond the capabilities of a new technology to the challenges it poses for human judgment. Mass-produced content that closely resembles reality tests people’s ability to distinguish the real from the fabricated. Platforms that customize each person’s viewing environment challenge the basis for shared facts, while increasingly sophisticated simulacra and intelligent spectacles make it harder to avoid passively inhabiting a world of images assembled by algorithms. These concerns extend beyond media to cognition, society, and civilization itself. Confronting this intelligent spectacle requires safeguarding the distinction between reality and fiction, the public’s capacity for independent judgment, and the autonomy of individuals.

Li Hong is an associate professor from the School of Journalism and Communication at Zhengzhou University.