AI Isn't Stealing My Creativity. It's Unlocking It: Sourced Research Brief, Script, and Production Demonstration

Full report

On this page

Section 1: Executive Summary and Theoretical Framework of Human Creativity

The emergence of generative artificial intelligence has ignited a fundamental debate across the creative industries. Critics characterize algorithmic synthesis as an existential threat to creative labor, arguing that large-scale generative models plagiarize human expression and induce cognitive atrophy. Conversely, technologists frequently frame generative models as autonomous creative agents capable of replacing human ideation. Both perspectives suffer from a reductive misunderstanding of creativity itself. Creativity is not a monolithic, instantaneous act of synthesis; it is an iterative, multi-stage cognitive and technical framework encompassing distinct phases: imagination, experimentation, selection, revision, storytelling, and executive artistic judgment1. To evaluate how artificial intelligence impacts human creative output, creativity must be contextualized as a teleological process—an intentional, goal-directed endeavor driven by lived experience, emotional resonance, and personal meaning3. Generative AI models operate through statistical pattern matching and reconstruction error minimization across vast high-dimensional latent spaces4. They possess neither intentionality nor consciousness; they generate probabilistic distributions of tokens or pixels based on prior training data4. Consequently, generative AI functions not as an autonomous author, but as a high-velocity feedback catalyst within the creative workflow2. In a human creative cognitive loop, the process originates in human lived experience and intent. This intent is translated into raw ideation and initial prompting, which prompts algorithmic synthesis in the latent space. Rather than accepting raw machine outputs, the human creator evaluates the results through spatial and layer inpainting, passes the refined elements into executive curation within a non-linear editor, and continuously iterates until achieving a final conceptual artifact that satisfies the original teleological vision1. When integrated intentionally, generative AI shifts the locus of human effort from manual, repetitive technical execution to high-level conceptual direction, structural curation, and aesthetic refinement5. The fundamental creative act resides not in the automated rendering of raw imagery or text, but in the originating vision and the critical judgment required to shape probabilistic output into a coherent narrative1. The core thesis of this investigation is grounded in this distinction: the machine can assist in synthesizing the visual and auditory artifacts, but the reason those artifacts exist originates exclusively from human intent3.

Section 2: Historical Trajectories of Disruptive Creative Technologies

The cultural panic surrounding generative AI mirrors historical reactions to preceding technological disruptions in visual and performing arts. A historical analysis reveals a consistent pattern: whenever a new technology drastically reduces the friction of technical execution, established practitioners predict the death of art, only for the medium to expand its creative horizons and evolve higher levels of abstraction2.

Photography and the Liberation of Painting (1839)

The introduction of the daguerreotype by Louis Daguerre in 1839 sparked widespread alarm among visual artists. French painter Paul Delaroche famously declared, "From today, painting is dead." Prior to photography, Western visual art was heavily evaluated on its ability to faithfully mimic physical reality. Photography automated portraiture and landscape documentation, rendering pure representational skill economically redundant for many artisans. However, rather than destroying painting, the camera liberated it. Unburdened from the obligation of representational accuracy, painters pivoted toward conceptual, emotional, and stylistic exploration, directly catalyzing Impressionism, Cubism, Expressionism, and modern abstract art2. The camera did not eliminate human creativity; it elevated the artist's focus from mimetic replication to expressive interpretation2.

Synthesizers and Electronic Music (1960s–1980s)

In the mid-20th century, the advent of commercial voltage-controlled synthesizers—such as those developed by Robert Moog and Don Buchla—met fierce resistance from traditional musicians' unions. Critics argued that electronic oscillation was "unnatural," predicting that synthetic instruments would bankrupt traditional orchestras and homogenize musical performance. In several jurisdictions, radio stations attempted to restrict synthesized broadcasts. Instead of destroying music, synthesizers democratized sound design, giving rise to entirely new musical genres, including synthwave, ambient, techno, hip-hop, and modern cinematic scoring2. The synthesizer transformed the recording studio into an instrument, shifting musical authorship toward sound architecture, arrangement, and timbre manipulation.

Digital Non-Linear Editing and CGI: The Jurassic Park Paradigm (1990s)

The transition from analog film cutting and physical optical effects to digital Non-Linear Editing (NLE) and Computer-Generated Imagery (CGI) provides the most direct historical parallel to modern generative workflows7. During the pre-production of Steven Spielberg's Jurassic Park (1993), stop-motion animation legend Phil Tippett was preparing intricate go-motion dinosaur armatures7. When Industrial Light & Magic (ILM) visual effects supervisors Dennis Muren and Mark Dippé demonstrated a photorealistic CGI test of a Tyrannosaurus rex walking across a daylight field, Tippett turned to Spielberg and remarked, "I think I'm extinct"7. Spielberg was so struck by the phrase that he inserted it directly into the film's script as a line delivered by Jeff Goldblum's character, Dr. Ian Malcolm10. Tippett's reaction reflected a momentary fear of technological displacement, but the actual outcome demonstrated paradigm hybridization rather than extinction8. Recognizing that CGI software programmers lacked deep knowledge of animal locomotion, physical weight distribution, and dramatic creature performance, Spielberg retained Tippett as the film's Dinosaur Supervisor8. Tippett and his team invented the Digital Input Device (DID)—an articulated physical armature fitted with digital encoders8. By posing the physical DID, Tippett's animators directly drove the digital 3D models inside the computer, combining traditional spatial performance skills with advanced digital rendering8. This historical precedent demonstrates that paradigm shifts displace specific mechanical techniques rather than underlying creative judgment and domain expertise8.

Section 3: Empirical Literature Review: Productivity, Quality, Originality, and Homogenization

Recent empirical literature provides quantitative insights into how generative AI alters individual output, creative confidence, task efficiency, and group-level idea diversity.

The Individual-Collective Paradox

The empirical data establishes what researchers term the "Social Dilemma of Generative AI"13. At the individual level, generative access acts as an equalizer and productivity multiplier13. Less inherently creative individuals (measured via the Divergent Association Task) experience significant gains in novelty (+10.7%) and usefulness (+11.5%) when provided with AI-generated ideation anchors13. The software establishes a high baseline floor for structural quality, grammar, and thematic coherence13. However, at the population level, unsupervised adoption induces a collective homogenization effect13. Because base large language models and diffusion systems sample from central tendencies within their training distributions, unrefined outputs converge toward normative, predictable stylistic modes13. In the Doshi & Hauser study, stories created with AI assistance demonstrated a 10.7% increase in semantic similarity compared to human-only controls13. This paradox highlights the critical role of human direction1. When creators treat generative tools as single-click solution engines, output flattens into generic tropes13. Conversely, when creators utilize AI strictly for rapid prototyping while exerting manual control over selection, revision, spatial composition, and editing, the individual quality boost is retained without succumbing to collective homogenization6.

The Jagged Technological Frontier and Cognitive Modes

Dell'Acqua et al. conceptualized the "Jagged Technological Frontier" to describe how AI capabilities do not scale predictably along human intuitive axes of task difficulty16. Tasks that appear visually or linguistically complex (e.g., rendering photorealistic volumetric lighting or synthesizing complex poetic meter) fall comfortably inside the frontier16. Conversely, tasks that seem trivial to human intuition (e.g., maintaining temporal consistency of fine anatomical details across shot cuts or adhering to exact physical spatial logic) fall outside the frontier, resulting in sudden, unpredictable model failures16. The study identified two dominant modes of successful human-AI integration20:

1. Centaurs: Creators who maintain task division, delegating discrete sub-tasks to the AI (e.g., generating texture maps) while reserving conceptual design and structural evaluation strictly for the human20.

2. Cyborgs: Creators who engage in continuous interaction with the model at the micro-task level, constantly prompt-adjusting, inpainting, and blending human inputs with machine responses20.

Both integration modes prevent the performance degradation observed when users blindly trust unverified model outputs19.

Section 4: Distinguishing Passive Generation from Intentional Human Direction

The debate surrounding AI art frequently collapses the distinction between passive prompt consumption and intentional directorial control1. To establish authorship, a workflow must incorporate human intervention, iterative revision, and selective filtering1. In a passive, single-click generative workflow, a user inputs a vague text prompt into an unconstrained model, receiving a contextless, unedited output. The machine dictates composition, lighting, style, and narrative pacing, resulting in homogenized visuals that reflect latent space averages13. In an intentional directional workflow, the human creator establishes a clear original concept and applies mindful friction before interacting with software15. The artist enforces structural spatial constraints using depth mapping or pose estimation masks, isolates specific regions for iterative inpainting, and composites raw generations inside a non-linear editor6. The creator applies manual color grading, precise audio layering, and temporal pacing cuts, ensuring every element serves a human narrative vision1.

Key Operational Components of Intentional Direction

• Mindful Friction: As Wharton research emphasizes, creators must establish "mindful friction" by completing preliminary conceptualizing, worldbuilding, and thematic structural framing before invoking generative tools15. Early reliance on raw prompts anchors the creator's imagination to central latent space tropes15.
• Spatial and Structural Constraints: Utilizing ControlNet models, depth maps, pose-estimation skeletons, and segmentation masks forces the AI to conform to exact camera focal lengths, actor positions, and spatial geometry dictated by the human director20.
• Iterative Inpainting and Layering: Rather than accepting a complete image generation, intentional creators isolate specific image regions (inpainting), adjusting micro-elements, background elements, and lighting interactions individually.
• Cinematic Non-Linear Assembly: Raw generative visual fragments are contextless. Human artistic judgment establishes cinematic rhythm, pacing, shot-to-shot spatial continuity, narrative pacing, color grading, and diegetic sound design within traditional editing suites6.

Section 5: Step-by-Step Production Demonstration: "The Memory Archivist of Neptune"

To demonstrate the progression of an original human idea through an AI-assisted pipeline into a finished short video, the following case study tracks the production of a 60-second sci-fi narrative titled "The Memory Archivist of Neptune."

In Stage 1, raw prompt outputs generated predictable action tropes featuring neon corridors and floating holograms. By applying mindful friction and custom system prompts, the narrative focus was redirected toward an emotional core: isolation versus preserved human warmth15. In Stage 2, spatial composition was established not through text prompting, but via a hand-drawn pencil blueprint converted into a ControlNet depth map20. This enforced rigid low-angle geometry, placing the archivist on the left third and an atmospheric window on the right third, preventing the AI from selecting a generic centered composition20. In Stage 3, the flat console rendered by the base diffusion model was masked and inpainted across multiple regional passes. Iterative text prompts introduced weathered brass valves, analog glass pressure gauges, and frost condensation. Flawed outputs—such as misaligned valves or floating glass—were discarded or manually composited in Photoshop before undergoing a low-denoising image-to-image pass to unify cast shadows and surface specular reflections. In Stage 4, motion synthesis presented temporal instability, illustrating the Jagged Technological Frontier16. Direct video generation produced unnatural facial warping during camera pushes19. The issue was resolved through a Centaur workflow: isolating the window frost animation pass, rotoscoping the static character from the keyframe, and compositing the elements in After Effects to achieve zero temporal distortion20. In Stage 5, individual visual assets were integrated into a professional Non-Linear Editing suite. The raw voice synthesis was manually time-stretched and pitch-adjusted in a Digital Audio Workstation, while generated music tracks were stripped of intrusive percussion to prioritize diegetic sub-bass hums and glass clinks. The final sequence reflects human directorial judgment across pacing, color balance, and sound design1.

Section 6: Complete 8–10 Minute YouTube Video Script

Title: AI Isn't Stealing My Creativity. It's Unlocking It. Estimated Run Time: 08:30 – 10:00 Format: Single-Presenter Video Essay / Visual Demonstration

[00:00 – 01:15] SECTION 1: THE EXTINCTION HOAX

Visual Cues: Fast-paced montage of sensationalist media headlines: "AI Will Replace Filmmakers," "The End of Human Art," "Is Photography Dead Again?" Cut abruptly to Presenter sitting in a warm, cinematic studio environment. Soft ambient lighting. On the desk: an old 35mm film camera, a mechanical keyboard, and a drawing tablet. Audio Cues: Tense, building orchestral synth drone that cuts out sharply when the Presenter speaks. Presenter: In 1993, during the pre-production of Jurassic Park, legendary stop-motion animator Phil Tippett sat in a dark screening room at Industrial Light & Magic7. He was watching an early CGI test of a Tyrannosaurus rex walking across a daylight field7. It was photorealistic, smooth, and unlike anything ever accomplished with physical puppets8. Tippett turned to director Steven Spielberg and said five words that would make cinematic history: "I think I'm extinct." \[cite: 7, 8\] Spielberg loved the line so much he put it right into the movie10. Jeff Goldblum says it to Sam Neill when they first see the living dinosaurs10. Visual Cues: B-roll clip of Jurassic Park (1993) – Ian Malcolm saying, "I think I'm extinct." Cut to archival photos of Phil Tippett working on the Digital Input Device armatures8. Presenter: It’s a great movie moment. But here’s the part most people forget: Phil Tippett didn’t go extinct8. Spielberg realized that while the computer programmers knew how to render pixels, they had no idea how an eight-ton predator actually moves, balances, or hunts8. He hired Tippett to supervise the dinosaur animations8. Tippett’s team invented an input armature—a physical puppet rigged with digital sensors—allowing stop-motion animators to drive digital 3D models using their hands8. The medium changed. The manual labor shifted. But the artistic judgment—the human understanding of performance and weight—remained indispensable8. Today, we are sitting in that exact same dark screening room. Looking at generative AI, millions of creators are staring at the screen, saying, "I think I'm extinct." They're wrong. The machine can help create the images, but the reason those images need to exist in the first place? That can only come from you3. On-Screen Graphic: TITLE CARD: AI ISN'T STEALING MY CREATIVITY. IT'S UNLOCKING IT.

[01:15 – 03:30] SECTION 2: DECONSTRUCTING THE MYTH OF ONE-CLICK ART

Visual Cues: Presenter transitions to screen-share mode. On one side, a simple prompt box. On the other side, a professional Non-Linear Editing timeline packed with video tracks, audio layers, color wheels, and mask nodes. Audio Cues: Upbeat, analytical lo-fi track begins under dialogue. Presenter: To understand why AI isn't going to replace creators, we have to stop treating "creativity" like a magical spark that happens in a single second1. Creativity is not typing five words into a text prompt, hitting generate, and taking a bow1. That isn't creation; that’s ordering fast food15. Real creativity is an iterative loop1. It’s imagination, experimentation, selection, revision, storytelling, and above all, artistic judgment1. Visual Cues: Screen capture showing a single prompt input: "Cinematic sci-fi movie scene, archivist on Neptune." The AI output pops up. It looks polished, but generic—filled with standard neon lights and floating holograms. Presenter: Look at this. This is a one-click generation15. It's glossy. It's photorealistic. And it is completely soulless13. Why? Because I gave up directional control1. I let the machine sample the average of its training data4. In 2024, researchers Anil Doshi and Oliver Hauser published a study in Science Advances13. They tested nearly 300 writers and over 3,000 story evaluations13. What they found was a fascinating paradox: giving people AI-generated ideas made individual stories better written, more enjoyable, and more novel—especially for people who struggled with baseline creativity13. On-Screen Graphic: Summary of Doshi & Hauser (2024) data:

\[cite: 13\]

• Individual Story Writing Quality: \+37.2%13
• Story Enjoyment: \+37.5%13
• Individual Story Novelty: \+8.1%13
• Collective Diversity across group outputs: \-10.7% (Homogenization)

Presenter: Did you catch that last number? While individual stories got better, the stories collectively became 10.7% more similar to each other13. Doshi and Hauser called it a social dilemma: writers are individually better off, but collectively, a narrower scope of content is produced13. If you use AI as a one-click slot machine, you get homogenized outputs13. You get the average of the internet4. But if you treat AI as a directional tool—if you inject human friction, spatial constraints, and strict revision—the entire dynamic flips6.

[03:30 – 06:00] SECTION 3: THE DEMONSTRATION: FROM VISION TO SCREEN

Visual Cues: Presenter sits at drawing tablet. Split screen showing raw sketching on tablet translating into ControlNet spatial depth maps, followed by isolated inpainting passes. Audio Cues: Focused, rhythmic sound design with mechanical keyboard clattering and gentle synthesizer pulses. Presenter: Let me show you what intent looks like in an AI pipeline. I wanted to make a short cinematic sequence called "The Memory Archivist of Neptune." My visual concept wasn't "generic sci-fi." My concept was a solitary worker in a freezing underwater outpost cataloging extinct human memories preserved in warm amber vials. Cyan atmospheric pressure outside; amber warmth inside. If I type that into a base diffusion model, it gives me holograms and lasers. It misses the emotional intent15. So, step one: Mindful Friction15. Visual Cues: Presenter holds up a hand-drawn index card with crude pencil composition lines. Presenter: I draw the composition myself20. A low-angle shot. Archivist on the left third. Massive pressure window on the right. I feed this drawing into ControlNet as a depth map20. This locks the camera perspective. The machine isn't deciding where the camera goes; I am. Visual Cues: Screen capture showing diffusion output conforming to the pencil sketch lines. Then, Presenter uses a mask tool to highlight a console. Presenter: Pass one gives me the right geometry, but the lighting on the console looks like a plastic toy. I don't re-generate the whole image. I mask out the console—this is Inpainting. I prompt specifically for "weathered brass valves, analog glass pressure gauges, frost condensation." I run ten variations. Nine are unusable. One has the exact metallic corrosion I envisioned. I select it, splice it in, and clean up the edges manually. Visual Cues: Cut to video generation interface. Keyframe is loaded. Presenter sets strict camera vector controls (Pan Left: 0, Push In: \+1.2). Presenter: Now, motion synthesis. Generative video models love to warp human faces if you let them run free19. That’s the "Jagged Technological Frontier"—a concept coined by Harvard researchers16. AI can render complex lighting inside the frontier, but it fails at maintaining temporal anatomy outside the frontier16. So I operate like a Centaur20. I split the tasks20. I generate the background window movement separately, rotoscope my character's face, and lock them together in After Effects. Visual Cues: Timeline view showing video layer 1 (animated frost window), video layer 2 (rotoscoped character), color correction adjustment layer, and 6 audio tracks (sub-bass hum, glass clinks, tape hiss, voiceover). Presenter: Look at this timeline. There are seven distinct software tools here. There are hours of manual rotoscoping, color balancing, and audio mixing. Did the AI generate the raw pixels? Yes. But did the AI direct the scene? Did it choose the color palette? Did it edit the cadence of the cut? No. The machine helped create the images, but the reason I wanted those images to exist came from me3.

[06:00 – 07:45] SECTION 4: ADDRESSING THE CRITICS

Visual Cues: Presenter returns on camera. Soft, authoritative framing. Text overlays displaying common criticisms appear and are systematically checked off. Audio Cues: Subtle, contemplative piano melody. Presenter: Now, let's address the criticisms directly, because there are real concerns that deserve honest answers. On-Screen Graphic: CRITICISM 1: "It’s not real art because you didn’t draw every pixel." Presenter: When darkroom photographers started dodging and burning exposures in 1890, painters said the exact same thing2. When electronic synthesizers emerged in 1970, orchestral musicians claimed pressing keys on a Moog wasn't "real music"2. Art is not defined by how much manual physical suffering you endure to render an asset2. Art is defined by teleological intent—by emotional resonance, narrative structure, and deliberate human choices1. On-Screen Graphic: CRITICISM 2: "AI models steal from artists to build their latent spaces." Presenter: This is a vital ethical issue. Training datasets must respect creator rights, opt-outs, and fair compensation. But conceptually, human artists learn in a similar way4. We study centuries of art history, absorb styles, synthesize patterns, and build an internal mental latent space. The goal of using AI isn't to copy an individual artist's style; it's to combine disparate concepts into something genuinely new that was previously trapped behind technical or financial barriers2. On-Screen Graphic: CRITICISM 3: "AI makes everyone lazy and homogenizes culture." Presenter: If you use it lazily, yes\! The research proves it: unguided AI flattens creative diversity13. But that isn't a flaw in the technology; it's a flaw in how it's used6. If you rely on the machine for your intent, you get generic results15. If you force the machine to serve your intent, you unlock visual possibilities that used to require a massive Hollywood budget3.

[07:45 – 09:30] SECTION 5: THE FINAL FILM & THE FUTURE OF CREATIVITY

Visual Cues: The screen transitions to full screen 16:9 cinematic playback of the completed 60-second short film, "The Memory Archivist of Neptune." Rich cyan tones, pristine glass frost animation, warm amber memory vial glow, deep immersive audio mix. Audio Cues: The completed short film's audio plays in full high fidelity—atmospheric sub-bass drone, gentle mechanical clicks, emotional synthesized string swell, articulate voiceover narrative. Voiceover (In Short Film): "They thought memory was digital. They thought light could be stored in cold silicon forever. But out here, under forty thousand atmospheres of ice and sea, we learned the truth: a memory isn't just data. It's warmth. And warmth requires a container that remembers what it felt like to burn." Visual Cues: The short film finishes with a fade to black. Cut back to Presenter on camera in studio. Presenter: That 60-second clip would have taken an independent animator six months to render manually three years ago. I built it in two days. Not because I skipped the creative process, but because the technical friction was stripped away, allowing me to iterate at the speed of my imagination5. We are not going extinct8. The cameras didn't kill painting2. Synthesizers didn't kill music2. CGI didn't kill animators8. Tools change2. The focus of creative labor shifts6. But as long as there is a story you desperately need to tell, a visual world you can see in your mind's eye, and the artistic judgment to guide it to reality—creativity remains entirely human1. Stop fearing the tool. Pick up the reins, enforce your intent, and unlock what you were meant to build6. Visual Cues: End screen graphics. Subscribe button, link to research brief, case study blueprints, and prompt breakdown down below in description. Final fade out. Audio Cues: Warm, inspiring music track swells to a resolve, then fades out.

Section 7: Counter-Arguments and Critical Defense of Authorship and Originality

To maintain a rigorous scholarly defense of AI-augmented human authorship, creators must directly address major philosophical, legal, and aesthetic objections.

The Labor Theory of Art vs. Teleological Intentionality

A prominent critique posits that artistic value is tied to the physical time and manual labor expended in production. Under this view, automated image synthesis constitutes cheating because it bypasses manual brushwork or 3D modeling. This argument relies on a fallacy refuted by art history2. Marcel Duchamp's readymades (Fountain, 1917\) and 20th-century conceptual photography established that art resides in selection, framing, context, and intent rather than pure physical manufacturing2. Generative tools shift effort up the abstraction ladder6. The labor does not vanish; it transforms into iterative curation, structural prompting, spatial composition, and temporal editing1.

The Risk of Cognitive and Skill Atrophy

Opponents argue that delegating ideation and rendering to automated systems will atrophy core human creative faculties, resulting in creators incapable of independent visual or narrative thought. Empirical evidence demonstrates that generative tools act as skill multipliers rather than replacements5. As Noy & Zhang demonstrated, automation of low-level drafting frees cognitive capacity for high-level editing, macro-structural revision, and conceptual refinement23. The artist's focus shifts from mechanical execution to executive critique—demanding higher aesthetic literacy and discernment6.

Overcoming Latent Space Homogenization

As established by Doshi & Hauser, default generative outputs converge toward generic modal representations13. Skeptics argue this inherent property makes genuine algorithmic originality impossible13. Model homogenization is the signature of passive, un-directed usage13. Diffusion models sample from central tendencies only when given high-level, under-specified text prompts4. When human directors apply specific spatial constraints (ControlNet), regional masking, custom image-to-image inputs, and post-production color and sound assembly, they force the model out of its default modes into unique aesthetic coordinates15.

Section 8: Conclusions and Strategic Recommendations

The integration of generative artificial intelligence into creative workflows marks a structural shift in human artistic expression, comparable to the adoption of photography, digital non-linear editing, and computer-generated imagery2. Empirical literature confirms that generative tools offer substantial benefits at the individual level—boosting task speed, raising baseline output quality, and democratizing visual production13. However, widespread, un-directed adoption presents a risk of collective homogenization, flattening cultural output into predictable latent space averages13. Navigating this technological shift requires rejecting both techno-dread and passive automation15. True creative empowerment resides in establishing human-centric workflows15:

• Retain Originating Intent: Human creators must establish conceptual, structural, and thematic frameworks before invoking algorithmic tools15.
• Practice Active Direction: Creators should employ spatial masking, depth mapping, regional inpainting, and manual post-production assembly to enforce directional control over model outputs20.
• Cultivate Executive Discernment: As technical execution becomes friction-free, competitive differentiation shifts to artistic judgment, critical selection, and storytelling coherence1.

Generative models provide powerful synthesis capabilities, but they remain tools incapable of human experience, emotional necessity, or personal intent3. The machine can assist in rendering the visual and auditory landscape, but the underlying drive to create—and the narrative purpose behind that creation—remains entirely human3.

Works cited

1. \[PDF\] Generative AI enhances individual creativity but reduces the, https\://www\.semanticscholar.org/paper/Generative-AI-enhances-individual-creativity-but-of-Doshi-Hauser/b3f40258164f0835c6683946dab16a953790289f

2. Art and The Science of Generative AI | PDF | Artificial Intelligence, https\://www\.scribd.com/document/790887726/Art-and-the-science-of-generative-AI-Science

3. If art is how we express our humanity, where does AI fit in? | MIT News, https\://news.mit.edu/2023/generative-ai-art-expression-0615

4. Art and the science of generative AI: A deeper dive \- arXiv, https\://arxiv.org/pdf/2306.04141

5. Generative AI and Creativity: A Systematic Literature Review and, https\://www\.researchgate.net/publication/392085967\_Generative\_AI\_and\_Creativity\_A\_Systematic\_Literature\_Review\_and\_Meta-Analysis

6. Does AI Limit Our Creativity? \- Knowledge at Wharton, https\://knowledge.wharton.upenn.edu/article/does-ai-limit-our-creativity/

7. 1 Jurassic Park Line Has An Incredibly Tragic Origin \- ScreenRant, https\://screenrant.com/jurassic-park-line-tragic-extinct-special-effects/

8. How Did They Get the CGI in the 1993 Jurassic Park to Look So, https\://www\.todayifoundout.com/index.php/2020/04/how-did-they-get-the-cgi-in-the-1993-jurassic-park-to-look-so-good/

9. Phil Tippett: Following his Imagination to the Stars and Beyond, https\://vfxvoice.com/phil-tippett-following-his-imagination-to-the-stars-and-beyond/

10. When Jurassic Park came out, did people fear the computer ... \- Reddit, https\://www\.reddit.com/r/vfx/comments/1mhufce/when\_jurassic\_park\_came\_out\_did\_people\_fear\_the/

11. TIL while directing Jurassic Park, when Spielberg saw CGI of T. rex, https\://www\.reddit.com/r/JurassicPark/comments/2g9dkv/til\_while\_directing\_jurassic\_park\_when\_spielberg/

12. Just in case the CGI for Jurrasic Park (1993) didn't live up to ... \- Reddit, https\://www\.reddit.com/r/MovieDetails/comments/f54kmw/just\_in\_case\_the\_cgi\_for\_jurrasic\_park\_1993\_didnt/

13. The Homogenization Engine: How LLMs Are Shrinking Cognitive, https\://brainbyteslab.org/articles/ai-homogenizes-expression-cognitive-diversity

14. (PDF) Generative AI enhances individual creativity but reduces the, https\://www\.researchgate.net/publication/382217365\_Generative\_AI\_enhances\_individual\_creativity\_but\_reduces\_the\_collective\_diversity\_of\_novel\_content

15. How AI Shapes Creativity: Expanding Potential or Narrowing, https\://ai.wharton.upenn.edu/updates/how-ai-shapes-creativity-expanding-potential-or-narrowing-possibilities/

16. dell-acqua-et-al-2026-navigating-the-jagged-technological, https\://www\.hbs.edu/ris/Publication%20Files/dell-acqua-et-al-2026-navigating-the-jagged-technological-frontier\_5c589c8c-fbb5-458f-b285-c944746cd717.pdf

17. Navigating the Jagged Technological Frontier: Field Experimental, https\://www\.hbs.edu/faculty/Pages/item.aspx?num=64700

18. Navigating the Jagged Technological Frontier: Field Experimental, https\://a-mcc.eu/en/library/articles-and-papers/navigating-the-jagged-technological-frontier-field-experimental-evidence-of-the-effects-of-ai-on-knowledge-worker-productivity-and-quality/

19. What Is the Jagged Frontier? Why AI Capabilities Are Smoothing Out, https\://www\.mindstudio.ai/blog/what-is-the-jagged-frontier-ai-capabilities

20. Navigating the Jagged Technological Frontier, https\://aiinstitute.hbs.edu/navigating-the-jagged-technological-frontier/

21. AI and the future of work: A tale about centaurs and cyborgs, https\://www\.siili.com/newsandinsights/ai-future-of-work-centaurs-and-cyborgs

22. A Nuanced Perspective on Harvard Business School's Jagged, https\://www\.lokad.com/blog/2024/4/8/a-nuanced-perspective-on-jagged-technological-frontier/

23. The productivity effects of generative artificial intelligence \- CEPR, https\://cepr.org/voxeu/columns/productivity-effects-generative-artificial-intelligence

24. Experimental evidence on the productivity effects of generative, https\://www\.researchgate.net/publication/372342901\_Experimental\_evidence\_on\_the\_productivity\_effects\_of\_generative\_artificial\_intelligence

25. Experimental evidence on the productivity effects of generative, https\://scispace.com/papers/experimental-evidence-on-the-productivity-effects-of-1km9slzy

26. The Projected Impact of Generative AI on Future Productivity Growth, https\://budgetmodel.wharton.upenn.edu/p/2025-09-08-the-projected-impact-of-generative-ai-on-future-productivity-growth/

27. \[2306.04141\] Art and the science of generative AI: A deeper dive, https\://arxiv.org/abs/2306.04141

28. Artificial Creativity: from predictive AI to Generative System 3, https\://www\.frontiersin.org/journals/artificial-intelligence/articles/10.3389/frai.2025.1654716/pdf

29. Harvard Business School Partners with BCG on AI Productivity Study, https\://www\.thecrimson.com/article/2023/10/13/jagged-edge-ai-bcg/