Art in the age of machine intelligence | Refik Anadol

Overview

Refik Anadol (media artist) describes his approach to AI-assisted generative art, treating data as a “pigment” and using architectural spaces as immersive canvases. In collaboration with machines, he aims for buildings and installations to feel like they can “dream” or “hallucinate.”

He frames the work around an idea that AI’s intelligence must be human-directed, and he repeatedly returns to questions about memory, imagination, and machine learning—including whether machines can dream, hallucinate, and connect experiences across people and time.

Artistic techniques, concepts, and creative processes

• Data-as-material (“data pigment”) Treating datasets as raw artistic substance that can be transformed into visual or experiential media.

• AI + generative algorithms Using machine learning to generate new forms, images, connections, and narratives from data.

• Architectural immersion Projecting or embedding artworks into real built environments so audiences experience data physically and collectively.

• Poetics of data Translating abstract datasets into sensory, poetic experiences rather than purely informational visuals.

• Memory as malleable interpretation Portraying remembering as changing or “melting” rather than static retrieval.

• Dreaming/hallucination as method Using systems that predict, synthesize, or reimagine outputs to create “dreamlike” fusion of past and future.

Projects and what they involve

“Virtual Depictions” (San Francisco)

• A public data sculpture depicting a fluid network of city connections.
• Makes invisible everyday data (e.g., social media feeds) visible and experientially “collective.”

“Wind-data paintings” (wind visualized as generative “poems”)

• Uses wind sensor datasets (speed, gust, direction).
• Converts measurements into an ethereal visual pigment via generative algorithms.
• Produces a meditative/speculative experience.

“Bosphorus” (kinetic data sculpture)

• Collects sea-surface data using high-frequency radar from the Marmara Sea.
• Projects the sea’s dynamic movement through machine intelligence.
• Creates an immersive environment that is calm but constantly changing.

“Archive Dreaming” (AI-driven public installation)

• Uses an open-source archive of ~1.7 million cultural documents spanning 270 years.
• Inspired by Borges’ “The Library of Babel.”
• Builds a user-driven immersive space for exploring vast archives through AI.

“Machine Hallucination” (NYC archive-based immersive project)

• Processes 100+ million city photographs using machine-learning algorithms.
• Uses a narrative system to generate hallucinated/predicted images.
• Blends past and future New York into a dreamlike experience.

“Melting Memories” (EEG/memory visualization)

• Triggered by Anadol’s uncle’s Alzheimer’s diagnosis.
• Collaborates with the Neuroscape Laboratory (UC) to interpret brain signals (EEG) related to memory formation.
• Visualizes remembering as melting/changing rather than disappearing.

LA Phil / Walt Disney Concert Hall installation (2018, 100-year symphony anniversary)

• Raises the question of whether a building can learn and dream.
• Uses 77 terabytes from archived LA Phil and WDCH recordings (spanning 100 years).
• Projects the archive onto the building’s skin using 42 projectors.

Immersive “data universe” TED Talk experience (near the end)

• Uses TED Talk datasets from the past 30 years: 7,705 talks and 7.4 million seconds.
• Processes 487,000 sentences into 330 topic clusters, including themes such as: nature, emissions, extinction, race issues, computation, trust, emotions, water, and refugees.
• Connects clusters algorithmically to generate 113 million line segments, revealing conceptual relationships.

Steps / materials / advice

• No explicit step-by-step “how-to” or materials list is provided.
• The process is described conceptually through the artworks: collect datasets → transform with generative AI → project/render into immersive spaces.

Creators, contributors, and institutions mentioned

• Refik Anadol — presenter/artist
• Jorge Luis Borges — inspiration for “The Library of Babel”
• Ridley Scott — mentioned via Blade Runner
• Frank Gehry — designer of Walt Disney Concert Hall (inspiration)
• Walt Disney Concert Hall (WDCH) / LA Philharmonic — archival sources for the symphony anniversary work
• Neuroscape Laboratory at the University of California — scientific collaboration for EEG/brain signal work
• TED — source of the referenced TED Talks dataset
• Additional collaborators referenced generally: architects, computer and data scientists, neuroscientists, musicians, storytellers (no further individual names provided).