When “Obligation” Replaces “Curiosity”: What is the Strategic Positioning of AI in the Creative Process?
The answer is straightforward: AI is transitioning from an auxiliary tool to a strategic-level variable reshaping narrative possibilities and economic models. Soderbergh’s “obligation” remark precisely highlights the anxiety of industry frontrunners—in the early stages of a technological paradigm shift, the greatest risk is not using the wrong tool, but being completely absent from the process of understanding it. This aligns with the logic of his earlier use of iPhones to make films: exploring the boundaries of tools to acquire narrative grammar and cost structures not yet mastered by competitors.
We can deconstruct this shift in “strategic positioning” from three levels:
- Narrative Level: AI is no longer just for special effects but intervenes in “representation” itself. In the case of the John Lennon documentary, faced with potentially fragmented, low-quality historical audio-visual archives, AI’s capabilities in restoration, frame interpolation, and even contextual reconstruction directly concern “what we can let the audience see and feel.” This touches on the core ethics of documentary filmmaking—the trade-off between authenticity and accessibility.
- Economic Level: According to internal assessments by the Motion Picture Association (MPA), visual effects and post-production costs already account for 25%-35% of the total budget of a mid-sized production film. The introduction of AI tools directly targets compressing this increasingly bloated cost. A more realistic projection is that within the next three years, approximately 15%-20% of routine visual effects work (such as object removal, simple scene extension, basic color grading) will be standardized and automated by AI tools.
- Process Level: AI is rewriting the linear workflow from pre-production to post-production. Directors can now use text to generate continuous concept art or dynamic storyboards during the script stage, enabling precise communication with cinematography and art departments. This not only speeds up decision-making but may also foster creativity that was previously unattainable due to technical or cost constraints.
The table below compares changes in key film and television production stages before and after AI intervention:
| Production Stage | Core Challenges in Traditional Mode | Potential Changes with AI Intervention | Current Technology Maturity (1-5) |
|---|---|---|---|
| Pre-Visualization | Highly reliant on concept artists, time-consuming communication, high modification costs. | Real-time generation of multiple versions of concept art and mood boards from text/sketches, accelerating creative iteration. | 4 |
| Visual Effects | Labor-intensive, long hours, extremely high computational costs for complex simulations (e.g., fluids, hair). | AI generates basic elements, automatic rotoscoping, intelligent scene repair, freeing artists to handle higher-level creative tasks. | 3 |
| Image Restoration & Enhancement | Restoring old films requires frame-by-frame manual processing, taking months or even years. | AI automatically removes noise, interpolates frames, colorizes, and upscales resolution, shortening restoration cycles by orders of magnitude. | 4 |
| Sound Design & Reconstruction | Complex foley and ambient sound recording/production, difficult historical audio restoration. | AI separates and cleans audio tracks, generates specific ambient sounds, and even simulates vocal characteristics of deceased actors (highly controversial). | 3 |
| Editing & Rhythm Analysis | Relies on editor experience and intuition, time-consuming footage screening. | AI analyzes scripts and all shot footage, marks emotional nodes, recommends edit points, and provides structural suggestions. | 2 |
mindmap
root(AI's Strategic Positioning Evolution in the Film Industry)
(Narrative Possibility Boundaries)
"Visualization" restoration and reconstruction of historical footage
Creating visual language beyond physical filming limitations
Generative potential for personalized narrative versions
(Production Economic Model)
Compressing post-production VFX and restoration costs
Accelerating pre-production development and decision cycles
Enabling ultra-low-cost/personalized production era
(Industry Power & Workflow)
Tool democratization vs. platform centralization
Creative decision-making shifts forward (director/writer)
Skill redefinition for post-production technical roles
(New Ethical & Legal Frontiers)
Blurred line between authenticity and "deepfakes"
Copyright ownership disputes over training data
Permanent challenges of performers' digital likeness rightsHollywood’s “Adoption Curve”: Who is Embracing, Who is Resisting, and Why?
Hollywood’s attitude towards AI is far from monolithic; it is a spectrum woven from interests, aesthetic beliefs, and generational differences. Soderbergh calls his peers’ resistance “a privilege,” a provocative statement that reveals the existing rift.
The Embracing Camp is primarily driven by two forces: first, “technologist-auteur” directors like Soderbergh, who see tools as extensions of their creative language. Second, studio and streaming platform executives, who see a ruthless arithmetic problem. According to industry data cited by The Wall Street Journal, for a typical streaming platform series launched in 2025, if its audience “completion rate” falls below expectations, the resulting hidden costs (including marketing waste and subscriber churn risk) could amount to tens of millions of dollars. The application of AI in pre-production script and audience analysis is seen as a way to reduce this uncertainty.
The Resisting Camp has more diverse and emotional voices. The core group consists of the vast base-level practitioners—from storyboard artists and VFX artists to foley artists. Their fear is very concrete: being replaced. A 2025 survey by the Visual Effects Society (VES) showed that over 65% of its members believe AI will significantly reduce traditional VFX job numbers within five years. Additionally, creative core members like writers and actors are more concerned about aesthetic homogenization and the demise of authorship. They resist not the technology itself, but the logic behind it that “in the name of efficiency and data, crushes creative intuition and manual craftsmanship.”
The slope of this adoption curve will be determined by one key factor: when the output quality of AI tools can cross from ‘impressive’ to ‘seamless and emotionally resonant.’ Currently, experimental videos from projects like Google DeepMind are still criticized for poor object persistence, weird physical logic, and a plastic look. This quality gap is the temporary moat for resisters and the technical summit that embracers must conquer.
The Twilight of Copyright? When AI Training Data Becomes the Next Oil Field Battle
The reason Soderbergh’s project has sparked controversy far beyond its technical discussion is that it inevitably touches the most sensitive nerve behind AI’s rapid advance: copyright. AI models require massive amounts of data for training, and the vast majority of this data comes from unauthorized web scraping, including copyrighted films, music, books, and artworks.
This creates a fundamental industry paradox: Hollywood studios are actively investing in AI tools to cut costs, while their vast content libraries, which they rely on for survival, are the “free” nourishment for these AI tools to learn from. This is a dangerous game of self-cannibalization.
The legal battle lines have already been drawn. Lawsuits, including those by several prominent authors, directly accuse companies like OpenAI and Meta of large-scale infringement. The EU’s AI Act explicitly requires transparency regulations for generative AI training data. In the US, the Copyright Office’s stance is still evolving, but a 2025 policy brief hinted that AI outputs that are direct, unaltered copies of style may struggle to receive copyright protection.
Future negotiations may head towards a “licensing economy.” Imagine large film and television archives (like Warner Bros., Disney’s film libraries) or celebrities’ digital likenesses becoming strategic assets requiring paid licenses for AI training. This could give rise to new power centers, significantly increase the compliance costs of AI applications, thereby slowing its adoption rate and reshaping the industry’s profit distribution landscape.
timeline
title Key Timeline of AI and Copyright Controversy
section 2023-2024
Wave of Lawsuits : Collective lawsuits by writers, artists<br>against OpenAI, Stability AI, etc.
Policy Exploration : US Copyright Office releases<br>initial draft guidelines on generative AI copyright
section 2025
EU Legislation : *AI Act* comes into effect<br>requiring training data transparency
Industry Turmoil : Hollywood writers' and actors' strikes<br>AI usage norms become a core demand
section 2026 (Current)
Soderbergh Case : Mainstream director's high-profile use of AI<br>in historical figure documentary<br>controversy intensifies
Legal Turning Point : Multiple lawsuits enter critical hearings<br>defining the boundaries of "fair use"
section 2027+ (Future)
New Normal Forms : Possible emergence of a "training data license" market<br>or expansion of copyright collective management organizations
Tool Diversification : Coexistence of compliant AI tools with "clean training"<br>and tools operating in "gray areas"The Next Five Years: Is AI a Transitional Trend or an Irreversible Industry Infrastructure?
Soderbergh himself offers an interesting prediction: in five years, we might look back and think this was just “an interesting phase.” This is a typical creator’s intuition—maintaining clarity amidst technological hype. However, from an industry infrastructure perspective, once a trend starts, it’s hard to reverse.
AI will not completely replace human creators, but it will inevitably redefine the meaning of “creator” and their workflow. The future film and television content ecosystem may exhibit a more distinct two-tier structure:
- Upper Tier: High-concept, strong narrative, live-action experience. This will be the anchor for cinematic artistic value and commercial investment. AI’s role here is as a powerful auxiliary and efficiency tool, used to realize previously impossible ideas or optimize production processes. Its core value remains “human stories and performances.”
- Lower Tier: Masses of genre-specific, customized, visually-driven content. This includes niche short films, advertisements, game cutscenes, educational content, etc. AI will play a core productivity role here, allowing very small teams or even individuals to produce content meeting certain quality standards at extremely low cost. This will intensify competition in the attention economy but also unleash tremendous creative energy.
Ultimately, the destination of technology depends on whose goals it serves. If AI merely serves capital’s maximization of efficiency, it may lead to content impoverishment and homogenization, triggering continuous resistance. But if, as Soderbergh practices, it is wielded by creators with a strong authorial consciousness as a “new brush” to explore new dimensions of narrative, then it has the potential to open an unprecedented era of creation. The key issue lies not in the tool itself, but in the power structures and creative philosophies behind it. Soderbergh’s “obligation” is precisely a declaration of intent to take the helm before the wave fully hits.
Further Reading
- The Directors Guild of America (DGA) position paper and seminar notes on AI applications in film and television production: https://www.dga.org/News/PressReleases/2025/250611-Artificial-Intelligence-Report.aspx
- Full text of the EU’s AI Act published in the Official Journal, see Chapter IV for provisions on general-purpose AI models and copyright: https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=OJ:L_202502269
- Visual Effects Society (VES) 2025 State of the Industry Report, containing survey data on AI’s impact on practitioners: https://www.visualeffectssociety.com/ves-state-of-the-industry-2025
