AI doppelgänger experiment – Part 1: The training
Generative AI that can quickly mimic an artist’s visual or writing style is raising hard questions about authorship, value, and copyright. Commenters argue that while AI easily reproduces commoditized, highly consistent styles used in commercial work, it still struggles with the deeper ideas, evolving aesthetics, and physical presence that define “great” art—yet this offers little comfort to freelancers and new artists whose livelihoods may be undercut. The debate centers on whether and how to protect artistic “style” without stifling human creativity, how much AI truly rivals human originality, and whether AI ultimately becomes a neutral tool, a parasitic threat, or even a new kind of artistic “master.”
Nature of Artistic Style and “Greatness”
- Several argue that artists who fixate on a narrow, commodified style are easy to replace; true “masters” continually change rules and can’t be predicted or cloned.
- Others counter that even highly innovative artists rely on recognizable periods or styles that can be mimicked once enough examples exist, so they are not immune.
- Debate over whether distinct modern styles (e.g., Cubism, Pollock-like work) are in fact easily reproducible by AI.
Economic Impact and Commoditization
- Many see the main threat to commercial illustrators, stock art creators, gig/freelance artists (e.g., Fiverr, tattoo designs, fandom niches), not high-end fine art.
- Some say artists already commodified themselves by selling repeatable styles; AI just accelerates that commodification.
- Concerns that emerging artists will be “kneecapped” as a few dozen public works may suffice to clone a sellable style before they build a career.
Ethics, Copyright, and Style Protection
- Common view: “style” is not and should not be copyrightable; trying to protect it legally risks chilling all artistic borrowing.
- Others want protections against training on specific artworks or using an artist’s name as a style brand, perhaps via trademark.
- Skepticism that technical defenses (Glaze/Nightshade, adversarial noise) can meaningfully prevent training.
- Some note simple workarounds like hiring copyists to train a “clean-room” clone of a style.
Comparisons to Past Technologies
- Parallels drawn to photography, film transitions, and digital tools: prior tech displaced specialists but didn’t kill art.
- Counterargument: AI is “parasitic” on existing media and requires ongoing human output to improve, unlike cameras.
Quality, Creativity, and Meaning
- Practitioners report AI can easily capture superficial style but struggles with deep concept, intent, and precise communication needs.
- Others claim AI is already traversing a “latent space of styles” and that human creativity is just points on a curve; some see this as philosophically bleak.
- Strong pushback that AI output still lacks the depth, surprise, and lived context of human art; accusations that its loudest boosters misunderstand art.
AI as Tool and New Medium
- Many artists use generative models for exploration, thumbnails, and style play, seeing AI as a powerful but dependent tool.
- Some predict a swing back to physical, unscannable, or experiential art as a refuge from commoditized digital imagery.
- There is interest in personal “doppelganger” models (for art and writing), but concerns about compute cost and ambivalence toward the broader AI ecosystem.