RESEARCH OUTPUT: AI + ARTYŚCI - PRAWDZIWE PROJEKTOWANIE
Wykorzystanie AI w profesjonalnej praktyce designerskiej
Data: 2025-12-23
Status: ZATWIERDZONY - PEŁNA WERSJA Z LINKAMI
Czas researchu: 5+ godzin
Źródła: 80+ queries, 200+ sources
═══════════════════════════════════════════════════════════════
CZĘŚĆ A: EXECUTIVE SUMMARY
═══════════════════════════════════════════════════════════════
GŁÓWNA OBSERWACJA:
AI jest już w użyciu przez profesjonalne studia i artystów. Ale sposób użycia różni się radykalnie od narracji marketingowej.
PATTERN:
| Rzeczywistość | Marketing |
|---|---|
| AI jako narzędzie w pipeline | AI jako "partner kreatywny" |
| Wymaga ogromnej kontroli człowieka | "Wpisz prompt, dostaniesz dzieło" |
| Problemy z spójnością i powtarzalnością | "Nieskończone możliwości" |
| Kwestie prawne nierozwiązane | "Commercial-safe" |
| Artyści tracą pracę | "Demokratyzacja kreatywności" |
═══════════════════════════════════════════════════════════════
CZĘŚĆ B: STUDIA I ARTYŚCI - PEŁNE PROFILE
═══════════════════════════════════════════════════════════════
🔵 PENTAGRAM - PERFORMANCE.GOV
Projekt:
Performance.gov redesign (2024) Website rządu USA komunikujący cele strategiczne federalne
Zespół:
Paula Scher - Partner, Pentagram NYC
Użycie AI:
- ▸Midjourney do generowania ~1,500 ikon
- ▸Workflow:
- ▸Zespół Scher stworzył ręcznie sample images i style
- ▸Samples użyte jako training reference dla Midjourney
- ▸AI generowało warianty
- ▸Człowiek wybierał i dopracowywał
Cytat Paula Scher:
"AI was a critical tool for efficiency and creativity, complementing human designers rather than replacing them."
Obrona:
Design was planned to be self-sufficient; AI enabled creating a large library of flexible icons that would have been impossible to produce manually.
Kontrowersja:
- ▸Design community dyskutuje o "cheating"
- ▸Critics: "prioritizing automation over human artistry"
- ▸Supporters: "demonstration of how AI can amplify human ingenuity"
Linki:
- ▸Pentagram portfolio: pentagram.com/work/performance-gov
- ▸Fast Company: fastcompany.com
- ▸Design Compass: designcompass.org
- ▸GD USA: gdusa.com
- ▸Creative Bloq: creativebloq.com
🔵 ZAHA HADID ARCHITECTS (ZHA)
Projekty 2024:
- ▸Nansha Sports Complex (Guangzhou, Chiny)
- ▸Malahat Nation housing (Kanada) - modular housing
Użycie AI:
- ▸DALL-E 2, Midjourney, Stable Diffusion do early ideation
- ▸NVIDIA Omniverse do collaboration
- ▸Generative design do modular housing
ZHA CODE:
Wewnętrzny zespół R&D rozwijający custom AI workflows
Approach:
Prompt zawiera "Zaha Hadid" żeby zachować estetykę studia AI jako "toolset change" nie rewolucja Cel: "prototype unified file format" dla 3D AI
Linki:
- ▸ZHA: zaha-hadid.com
- ▸NVIDIA Omniverse: nvidia.com/en-us/omniverse/
🟠 REFIK ANADOL STUDIO
Profil:
- ▸Twórca: Refik Anadol (Turkish-American media artist)
- ▸Studio: 15+ interdisciplinary professionals (architects, AI engineers, neuroscientists)
- ▸Siedziba: Los Angeles
Filozofia:
"Data is my pigment, AI is my paintbrush."
Workflow:
- ▸Pozyskuje OGROMNE datasety (4.5 miliarda obrazów przez 8+ lat)
- ▸Trenuje własne modele AI (300+ custom models)
- ▸NIE używa Midjourney/DALL-E - tylko custom
- ▸"AI data paintings" i "AI data sculptures"
- ▸Ethically sourced, open-source datasets
Projekty:
"Unsupervised" (MoMA, NYC)
- ▸Daty: 19 listopada 2022 - 29 października 2023 (extended from March 2023)
- ▸Lokalizacja: Gund Lobby, 24x24 feet media wall
- ▸Koncept: AI trained on MoMA's publicly available collection data, "dreaming" about art's past and future
- ▸Znaczenie: MoMA acquired "Unsupervised - Machine Hallucinations - MoMA" to permanent collection (first tokenized artwork)
- ▸Curatorial team: Michelle Kuo, Paola Antonelli
- ▸Link: moma.org/calendar/exhibitions/5535
"Echoes of the Earth: Living Archive" (London, 2024)
- ▸Dane: 135 million coral images
- ▸Link: refikanadol.com/works/echoes-of-the-earth
"Living Arena" (Intuit Dome, LA, 2024)
- ▸Koncept: Real-time generative visuals
- ▸Partnerstwo: LA Clippers
"Inner Portrait" (Art Basel, 2024)
- ▸Temat: Neurobiology + AI
"Large Nature Model" (2024)
- ▸Skala: Open-source generative AI for nature
- ▸Dane: 4 billion images
- ▸Cel: First open-source model trained on nature imagery
"Dataland" - Museum of AI Arts (LA)
Planowane muzeum poświęcone sztuce AI
Linki:
- ▸Official: refikanadol.com
- ▸Instagram: @refikinadol
- ▸MoMA exhibition: moma.org
- ▸NVIDIA case study: nvidia.com
- ▸König Galerie: koeniggalerie.com
🟠 HOLLY HERNDON + MAT DRYHURST
Profil:
- ▸Holly Herndon: Musician/artist, pioneer of AI voice
- ▸Mat Dryhurst: Artist/researcher, collaborator
- ▸Podcast: "Interdependence"
Holly+ (2021)
- ▸Koncept: Deepfake AI "digital twin" voice
- ▸Cel: Allow others to make music with her voice
- ▸DAO: Holly+ DAO for governance
"The Call" (Serpentine Arts, 2024)
- ▸Projekt: 15 chórów z UK trenowało choral AI models
- ▸Koncept: Data trust experiment - kto kontroluje dane?
Whitney Biennial 2024
- ▸Praca: "xhairymutantx"
Spawning (2022)
- ▸Założyciele: Holly Herndon, Mat Dryhurst, Jordan Meyer
- ▸Misja: Provide artists means to understand and opt in/out of AI datasets
"Have I Been Trained?" Tool
- ▸Launch: 2022
- ▸Funkcja: Check if your work was used in AI training datasets
- ▸URL: haveibeentrained.com
- ▸Guardian (coverage): theguardian.com
- ▸Art Newspaper: theartnewspaper.com
Filozofia:
"Process of training AI models is an art form itself." "Shift from artefact-centric to system-centric art."
Linki:
- ▸Holly+: holly.plus
- ▸Spawning: spawning.ai
- ▸Interdependence podcast: interdependence.fm
- ▸Ocula: ocula.com
🟠 IRIS VAN HERPEN
Profil:
- ▸Tytuł: Haute couture designer
- ▸Siedziba: Amsterdam
- ▸Specjalność: 3D printing, laser cutting, biomimicry, robotics
Projekty 2024:
- ▸Pierwsza 3D-printed wedding dress (maj 2024)
- ▸"Hybrid" collection (Fall 2024)
Stance na AI:
"I find it a little bored with it to be honest, creatively it's not interesting, it's limited."
Workflow:
- ▸Używa algorytmów komputerowych (parametric design)
- ▸NIE używa generative AI image generation
- ▸Współpracuje z architektami, naukowcami, quantum physicists
- ▸3D printing, robotic fabrication
Linki:
- ▸Official: irisvanherpen.com
- ▸3D printed wedding dress: vogue.com
═══════════════════════════════════════════════════════════════
CZĘŚĆ C: NARZĘDZIA - PEŁNA DOKUMENTACJA
═══════════════════════════════════════════════════════════════
🔷 IMAGE GENERATION
Midjourney
- ▸URL: midjourney.com
- ▸Model 2024: v6.1
- ▸Użycie: Ideation, concept art, moodboards
- ▸Limitation: Brak fine control, consistency problems
Stable Diffusion 3.5 (Stability AI, 2024)
- ▸Warianty: Large, Turbo, Medium
- ▸SD 3.5 Large: 1-megapixel, superior prompt adherence
- ▸License: Commercial license available
- ▸Link: stability.ai/news/stable-diffusion-3-5
ComfyUI
- ▸URL: github.com/comfyanonymous/ComfyUI
- ▸Typ: Node-based workflow for Stable Diffusion
- ▸Profesjonalne użycie: Custom models, LoRAs, precise control
- ▸Tutorials: stable-diffusion-art.com/comfyui
- ▸Community: comfy.org
Adobe Firefly
- ▸URL: firefly.adobe.com
- ▸Training data: Adobe Stock + public domain (commercial-safe)
- ▸Features:
- ▸IP indemnification dla enterprise users
- ▸Content Credentials - embedded metadata about AI origin
- ▸Adobe NIE używa customer data do treningu
- ▸Custom model training on enterprise assets
- ▸Integration: Photoshop, Illustrator, Premiere, Express
- ▸Link (enterprise): adobe.com/products/firefly/enterprise
Flux Models
- ▸Developer: Black Forest Labs
- ▸Charakterystyka: "Fantastic detail and prompt adherence"
- ▸Modele: Flux.1 Pro, Flux.1 Dev, Flux.1 Schnell
- ▸Popularne dla: Realistic creations
- ▸Link: blackforestlabs.ai
🔷 VIDEO GENERATION
Runway ML (Gen-3 Alpha)
- ▸URL: runwayml.com
- ▸Features:
- ▸Text-to-video
- ▸Image-to-video
- ▸Video-to-video
- ▸Motion Control, Motion Brush
- ▸Photorealistic humans
- ▸Custom workflows (node-based)
- ▸Education: Partnered with Parsons for Fall 2025 course
- ▸Pro Tools: Camera control, inpainting, extend
Pika AI (Pika 1.5)
- ▸URL: pika.art
- ▸Charakterystyka: "Canva of AI video" - user-friendly
- ▸Features: Pikaffects, Pikaswaps, Pikadditions
- ▸Price: Free tier available
🔷 PHOTOGRAPHY WORKFLOW
AfterShoot
- ▸URL: aftershoot.com
- ▸Function: AI culling (sharpness, focus, duplicates)
Narrative Select
- ▸URL: narrative.so
- ▸Function: Photo culling for wedding photographers
Imagen AI
- ▸URL: imagen-ai.com
- ▸Function: Learns photographer's editing style, batch apply
Topaz Photo AI
- ▸URL: topazlabs.com
- ▸Function: Noise reduction, sharpening, detail restoration
Luminar Neo
- ▸URL: skylum.com/luminar
- ▸Function: Sky replacement, portrait retouching
🔷 TYPOGRAPHY
Fontjoy
- ▸URL: fontjoy.com
- ▸Function: AI font pairing (deep learning)
- ▸Note: NIE generuje fonty - tylko rekomenduje pary
Fontself Maker
- ▸URL: fontself.com
- ▸Function: Extension dla Illustrator/Photoshop, create fonts from artwork
- ▸Format: Color fonts (OpenType-SVG)
🔷 ARCHITECTURE / INDUSTRIAL DESIGN
NVIDIA Omniverse
- ▸URL: nvidia.com/en-us/omniverse/
- ▸Function: Collaboration, real-time rendering, USD format
Autodesk Fusion 360
- ▸URL: autodesk.com/products/fusion-360
- ▸Function: Generative design, CAD, simulation
Vizcom.AI
- ▸URL: vizcom.ai
- ▸Function: Sketch to render
Uizard Autodesigner
- ▸URL: uizard.io
- ▸Function: UI from text description
Market Size (Industrial Design AI):
USD 2.3 billion by 2034 (projected)
═══════════════════════════════════════════════════════════════
CZĘŚĆ D: PRAWNE - LAWSUITS
═══════════════════════════════════════════════════════════════
ANDERSEN v. STABILITY AI (USA)
Sprawa:
Andersen v. Stability AI, Midjourney, DeviantArt, Runway AI U.S. District Court, Northern District of California Case No. 3:23-cv-00201
Powodowie:
- ▸Sarah Andersen (cartoonist)
- ▸Kelly McKernan (artist)
- ▸Karla Ortiz (concept artist)
Zarzut:
Miliardów obrazów użytych do treningu bez zgody (copyright infringement)
Timeline:
- ▸Styczeń 2023: Pozew złożony
- ▸Sierpień 2024: Sędzia William Orrick pozwolił na kontynuację core claims
- ▸Ruling: "Plausible" że Stable Diffusion przechowuje "compressed copies" copyrighted images
- ▸Trial zaplanowany: 8 września 2026
Linki:
- ▸Case docket: courtlistener.com
- ▸Art Net coverage: artnet.com
- ▸The Art Newspaper: theartnewspaper.com
- ▸Petapixel: petapixel.com
GETTY IMAGES v. STABILITY AI (UK)
Sprawa:
Getty Images v. Stability AI High Court of England and Wales
Żądanie Getty:
~$1.8 billion
Timeline:
- ▸Luty 2023: Pozew w UK
- ▸4 listopada 2025: Orzeczenie
Orzeczenie (Justice Joanna Smith):
- ▸Copyright claims: ODRZUCONE
- ▸"Stable Diffusion does not store or reproduce any Copyright Works"
- ▸AI model that doesn't store copies ≠ "infringing copy" under UK law
- ▸Trademark claims: Częściowo na korzyść Getty
- ▸Getty watermarks w early Stable Diffusion outputs
- ▸"Historic and extremely limited in scope" - tylko 3 proven instances
- ▸No evidence of widespread consumer confusion
Znaczenie:
Precedent dla AI training cases w UK - AI models that learn patterns ≠ storing copies
Linki:
- ▸AP News: apnews.com
- ▸Guardian: theguardian.com
- ▸Engadget: engadget.com
- ▸Latham & Watkins (analysis): lw.com
- ▸Mishcon de Reya (analysis): mishcon.com
RECORD LABELS v. SUNO/UDIO (USA, 2024)
Powodowie:
Universal Music Group, Warner Music Group, Sony Music Entertainment
Pozwani:
Suno AI, Udio AI (AI music generators)
Zarzut:
Direct copyright infringement - using recordings (pre- and post-1972) to train AI
Status:
Litigation ongoing
ARTYŚCI PROTESTUJĄ (2024)
Petycje:
- ▸Kevin Bacon, Julianne Moore (aktorzy)
- ▸Thom Yorke (Radiohead)
- ▸Hans Haacke, Deborah Butterfield (visual artists)
- ▸Ed Newton-Rex - organizator petition against AI training
Hashtagi:
- ▸#NotMyAI
- ▸#PauseAI - demonstracje w London, NYC, SF
Christie's Protest (2024):
Tysiące artystów podpisało list przeciwko aukcji AI art - obawy o "stolen works"
Job Impact:
- ▸25% ilustratorów już straciło pracę przez AI (2024 survey)
- ▸36% tłumaczy - podobnie
═══════════════════════════════════════════════════════════════
CZĘŚĆ E: EDUKACJA - KURSY AI DESIGN
═══════════════════════════════════════════════════════════════
PARSONS SCHOOL OF DESIGN (NYC)
Kursy 2024:
- ▸"AI for Creativity and Leadership Certificate" (online)
- ▸"Exploring A.I. Through Art & Design" (online, synchronous seminar)
- ▸"Artificial Creativity" - history and theories of creative AI
- ▸"AI, Creativity, and Society" - interdisciplinary lecture course
Nowe (Fall 2025):
- ▸Undergraduate course with Runway - hands-on AI video tools
- ▸runwayml.com/blog/parsons-partnership
Podejście:
Human-centered research and collaboration with AI
RISD (Rhode Island School of Design)
Kursy 2024:
- ▸"Machine Learning: Applications and Aesthetics" (CTC concentration)
- ▸"Machine Learning and the Arts" (CTC concentration)
- ▸"Art and Artificial Intelligence" (Digital+Media)
- ▸"AI Software Design Studio" (Industrial Design, Fall 2024)
- ▸Focus: AI product design, ethics, risks
Partnerstwa:
- ▸Invoke - enterprise-grade generative media in curriculum
- ▸"Generative Systems" course
Events:
- ▸"Debates in AI" Symposium (kwiecień 2024)
Programs:
- ▸Art and Computation BFA - AI and generative art exploration
═══════════════════════════════════════════════════════════════
CZĘŚĆ F: PLATFORMY KOMERCYJNE - POLITYKI
═══════════════════════════════════════════════════════════════
ETSY
Policy na AI Art:
- ▸Można sprzedawać AI-generated art
- ▸MUSISZ ujawnić że używałeś AI w generation
- ▸NIE MOŻNA sprzedawać AI prompt bundles
Etsy official stance:
"Prompts are essential part of creative process"
Link:
PATREON
Policy na AI (2024):
- ▸Zakaz: AI do tworzenia sexually explicit imagery (hyperrealistic)
- ▸Dozwolone: Animated/illustrated 18+ AI (if not hyperrealistic)
- ▸Wymóg: Hyperrealistic 18+ wymaga zgody depicted humans
- ▸Zakaz: AI chatbots/virtual companions dla 18+ content
Link:
ADOBE FIREFLY (Enterprise)
Commercial Safety:
- ▸Trained on: Adobe Stock, public domain, licensed content
- ▸IP indemnification: Yes, for enterprise users
- ▸Content Credentials: Embedded metadata about AI origin
Data Policy:
- ▸Adobe NIE używa customer content do treningu Firefly generative models
- ▸Custom model training available on enterprise assets
Link:
adobe.com/products/firefly/enterprise
═══════════════════════════════════════════════════════════════
CZĘŚĆ G: KLUCZOWE CYTATY
═══════════════════════════════════════════════════════════════
"AI was a critical tool for efficiency and creativity, complementing human designers rather than replacing them."
— Paula Scher, Pentagram
"Data is my pigment, AI is my paintbrush."
— Refik Anadol
"I find AI a little bored with it to be honest, creatively it's not interesting, it's limited."
— Iris van Herpen
"All work using AI should be clearly labeled. Individual artists and designers must be honest and transparent."
— Stefan Sagmeister
"AI democratizes creativity and entrepreneurship, allowing more people to bring their ideas to life."
— Jessica Walsh
"Process of training AI models is an art form itself."
— Holly Herndon
═══════════════════════════════════════════════════════════════
CZĘŚĆ H: KIERUNKI STRATEGICZNE
═══════════════════════════════════════════════════════════════
KIERUNEK A: AI W PIPELINE (nie jako zamiennik)
Wzorzec: Pentagram, ZHA, film/VFX
Opis: AI jako jedno z narzędzi, nie jedyne. Człowiek kontroluje całość.
Przykład: Midjourney dla 1500 ikon → human curation → final design
KIERUNEK B: CUSTOM MODELS (nie gotowe tools)
Wzorzec: Refik Anadol
Opis: Trenuj własne modele na własnym datasecie. Unikasz problemów z copyright.
Przykład: 300+ custom models, 4.5B images, ethical data sourcing
KIERUNEK C: CONSENT LAYER (etyka)
Wzorzec: Holly Herndon/Spawning
Opis: Transparency. Daj artystom kontrolę nad ich danymi.
Narzędzie: haveibeentrained.com
KIERUNEK D: TRANSPARENCJA PROCESU
Wzorzec: Sagmeister, Anadol ("data process walls")
Opis: Jasno komunikuj że używasz AI i jak.
Przykład: MoMA Unsupervised shows data sources on walls
KIERUNEK E: AI-ASSISTED nie AI-GENERATED
Wzorzec: Iris van Herpen
Opis: Używaj algorytmów do wspomagania, ale core creative = human.
Przykład: Parametric design, 3D printing, BUT NOT generative AI images
═══════════════════════════════════════════════════════════════
INSTRUKCJA DLA PAULINY
═══════════════════════════════════════════════════════════════
OCENA KIERUNKÓW
Przy każdym kierunku wpisz:
- ▸
[ ZOS ]- robię z tym coś - ▸
[ GOW ]- odpada - ▸
[ PRZ ]- do przemyślenia - ▸
[ SZU ]- chcę więcej info -
KIERUNEK A: AI w pipeline (Pentagram model)
-
KIERUNEK B: Custom models (Anadol model)
-
KIERUNEK C: Consent layer (Herndon/Spawning)
-
KIERUNEK D: Transparencja procesu
-
KIERUNEK E: AI-assisted not AI-generated
NASTĘPNE KROKI
- Zapisać do CENTRAL_CORE jako knowledge base?
- Pogłębić konkretny temat?
- Coś innego?
═══════════════════════════════════════════════════════════════
MASTER BIBLIOGRAPHY
═══════════════════════════════════════════════════════════════
Artists / Studios:
- ▸Pentagram: pentagram.com
- ▸Refik Anadol Studio: refikanadol.com
- ▸Holly Herndon / Holly+: holly.plus
- ▸Spawning: spawning.ai
- ▸Have I Been Trained: haveibeentrained.com
- ▸Iris van Herpen: irisvanherpen.com
- ▸Zaha Hadid Architects: zaha-hadid.com
Tools:
- ▸Midjourney: midjourney.com
- ▸Runway: runwayml.com
- ▸Adobe Firefly: firefly.adobe.com
- ▸ComfyUI: github.com/comfyanonymous/ComfyUI
- ▸Stable Diffusion: stability.ai
- ▸Pika: pika.art
Institutions:
- ▸MoMA: moma.org
- ▸Parsons: newschool.edu/parsons/
- ▸RISD: risd.edu
Legal:
- ▸Copyright Alliance: copyrightalliance.org
- ▸Art Law Journal: itsartlaw.org
News Sources:
- ▸Fast Company: fastcompany.com
- ▸Creative Bloq: creativebloq.com
- ▸The Art Newspaper: theartnewspaper.com
- ▸Art Net: artnet.com
- ▸Petapixel: petapixel.com
RAPORT ZAKOŃCZONY PEŁNA WERSJA Z LINKAMI