AI-generated films are emerging as a potential future of entertainment, but a significant gap exists between what the technology can produce and what audiences actually want to watch. Chinese video platform iQIYI announced in April 2026 that it is developing a fully AI-generated movie for its Ferryman series, partnering with GHY Culture & Media and establishing an AI talent pool with over a hundred celebrity likenesses. However, the path forward faces substantial obstacles, from technical limitations to fundamental questions about whether viewers will accept synthetic performances in place of human actors.

The timing of iQIYI's announcement comes just weeks after OpenAI's dramatic exit from the consumer AI video market. In late March 2026, OpenAI shut down Sora, its text-to-video generation tool that could create realistic videos up to 60 seconds long from text prompts. The closure included discontinuing Sora's standalone app, API interface, and its integration within ChatGPT. According to industry analysis, the shutdown reflected multiple challenges: high operational costs, difficulty monetizing AI video content, and strategic downsizing amid pressure from an initial public offering. The failure of Sora, despite its technical sophistication, signals that the market for AI-generated video content may be far smaller than early advocates predicted.

What Technical Problems Are Holding Back AI-Generated Films?

Even when AI video generation works as intended, the output frequently suffers from quality issues that undermine viewer immersion. These technical shortcomings stem from fundamental limitations in how AI models understand and render the physical world.

• Physical Logic Failures: AI lacks genuine understanding of causal relationships and physical laws such as gravity, collision, and occlusion, resulting in unrealistic or illogical outputs like food passing through teeth during eating scenes or objects merging into one another.
• Frame Continuity Issues: Within seconds, objects, characters, or backgrounds may mutate unexpectedly; clothing colors may suddenly change, limbs may appear abnormal with extra fingers or missing arms, and inconsistencies in time and space become apparent.
• Texture and Lighting Realism: Skin appears overly smooth and artificial, metals lack proper sheen, lighting fails to reflect environmental sources accurately, and depth rendering is poorly executed, creating a distinctly synthetic appearance.
• The Uncanny Valley Effect: Character movements tend to be stiff and unnatural, with vacant gazes, lack of dynamic muscle texture, and rigid movements that make AI-generated humans instantly recognizable as "fake," much like plastic flowers that look attractive from a distance but lack vitality up close.

These technical flaws accumulate to create a viewing experience that feels fundamentally inauthentic. Unlike human performances, which contain minor imperfections that form part of the creative process and artistic expression, AI-generated films resemble "copy-and-paste outputs" that lack genuine lived experience.

Why Are Actors Hesitant to License Their Likenesses?

iQIYI's CEO Gong Yu framed AI-generated films as a potential benefit to actors, suggesting they could license their likenesses and reduce their workload. The platform claims to have secured AI licensing agreements with over a hundred celebrities, including Chen Zheyuan, Zeng Shunxi, Cheng Lei, Jiang Long, Ma Su, and stand-up comedian Director Fang. However, several prominent actors have publicly denied involvement.

Zhang Ruoyun, Yu Hewei, Li Yitong, and others issued statements through their studios denying that they had signed any AI film and television authorization agreements. Their resistance reflects deeper concerns about reputation and authenticity. Actors have spent years building carefully crafted public personas through genuine performances, and many are unwilling to allow AI to replace, deconstruct, blur, distort, or damage those authentic images in the minds of audiences. The fear is not merely about lost work, but about losing control over how their likenesses are used and portrayed.

How Are Audiences Reacting to the Prospect of AI Films?

Beyond technical and ethical concerns, a fundamental question remains unanswered: will audiences actually watch AI-generated films? Early indicators suggest skepticism. Some viewers have already expressed doubts, asking rhetorically whether anyone will watch AI films if live-action productions already struggle to attract audiences. One particularly pointed observation suggested that if all actors are AI-generated, perhaps AI viewers should also be the ones paying for subscriptions.

Such remarks reflect deeper anxieties about authenticity and value. If actors contribute only their faces while all movements and special effects are handled by AI, the result may feel excessively artificial. Audiences recognize the difference between genuine human performance and synthetic recreation, and many appear unwilling to accept the latter as a substitute. The market will ultimately decide, but early signals from Sora's shutdown suggest that consumer appetite for AI-generated video content may be limited.

Steps to Understanding the Future of AI in Entertainment

• Monitor Technical Progress: Track improvements in AI video generation, particularly in addressing physical logic errors, frame continuity, and the uncanny valley effect, as these remain the primary barriers to viewer acceptance.
• Observe Market Response: Pay attention to how audiences respond to iQIYI's AI-generated Ferryman film and similar projects, as box office performance and subscription metrics will reveal whether the market can sustain AI-generated content.
• Follow Regulatory Developments: Watch for emerging regulations around AI likeness rights, deepfakes, and actor consent, as legal frameworks will shape how AI talent pools can operate.
• Assess Economic Viability: Examine whether AI-generated films can achieve profitability given production costs, licensing fees, and the apparent difficulty in monetizing AI video content that Sora's shutdown revealed.

The closure of Sora provides a cautionary precedent. Despite OpenAI's technical capabilities and resources, the company determined that the consumer AI video market was not viable. The difficulty of monetization suggests that fully AI-generated video content has not been well received by the public and has failed to capture meaningful market share. If a tool as sophisticated as Sora could not sustain commercial viability, the challenges facing iQIYI's AI film ambitions may be even more formidable.

The fundamental issue is one of authenticity and human connection. Film and television have always derived their power from the ability of human actors to convey emotion, vulnerability, and lived experience. When audiences watch a performance, they are witnessing a real person making real choices, even if those choices are scripted. AI-generated performances, no matter how technically advanced, lack this essential element of human authenticity. Until AI can convincingly replicate not just the appearance of human performance but the emotional truth beneath it, audiences may continue to prefer the imperfect reality of human actors over the polished artificiality of AI.

" }