AURORA Biennial 2026: How Artists Are Redefining Art and AI
Scheduled to take place from November 4 to 21 across Dallas and Fort Worth, the AURORA Biennial 2026 highlights how creators are leveraging accessible data streams and artificial intelligence to push the boundaries of artistic exploration.
Julia Kaganskiy, co-curator for the upcoming biennial, points out that the convergence of creative expression and technology has been evolving over the span of several decades.
Early pioneers such as Vera Molnár and Harold Cohen had to write their own software, build drawing machines or collaborate with engineers to create computer-generated art. Today, as computational tools become more accessible and barriers to entry fall, that expansion is accelerating.
Art Challenges How We See and Use Technology
Spanning disciplines from poetry and climate awareness to the nuances of riding in driverless transit, the exhibition features creators building interactive digital experiences powered by advanced technology.
Notable works include Sasha Stiles's A Living Poem, a generative language model trained on her own verse; LaJuné McMillian’s The Portal’s Keeper, an interactive AI project shaped by personal voice, text, and motion tracking; and Theo Triantafyllidis’s Drift Lattice, which employs real-time ocean and climate feeds to animate a living marine ecosystem.
Our Relationship With Autonomous Vehicles
Lauren Lee McCarthy’s installation, AUTO, utilizes a simulated passenger journey on a driverless bus to investigate public transit and automation. The piece prompts critical inquiries regarding how autonomous systems transform the passenger dynamic and the changing role of individuals in heavily automated environments.
Offering another angle on this subject, Anne Berens—CEO and CTO of SunLED, a firm specializing in human-centric near-infrared interior lighting—notes that while engineers successfully engineered Waymo vehicles to navigate autonomously, they may have overlooked how passengers actually spend their travel time.
Berens urges designers to rethink vehicle cabins completely from the vantage point of the rider.
I would strip everything from the interior and really start with an empty box, and think about, okay, when I am going from A to B and I don’t need to think about the driving activity itself, then what am I doing in this box? I’m either relaxing, or I’m connecting with people, or I’m working.
When Art Changes the Conversation
Trevor Paglen’s interactive exhibit, Faces of ImageNet, explores how human prejudice infiltrates machine-learning models through flawed training data.
Kaganskiy characterizes the installation as a direct continuation of ImageNet Roulette, a viral 2019 collaboration between Paglen and AI researcher Kate Crawford. That earlier experiment permitted visitors to upload selfies to observe how AI categorized them using Princeton’s 2009 ImageNet database.
Public backlash over offensive labels exposed deep-seated biases within machine learning. According to Kaganskiy, this pushed the ImageNet collective to purge roughly 600,000 images and 1,593 subcategories deemed unsafe or sensitive. Although problematic content has been scrubbed, Faces of ImageNet serves as a persistent warning regarding automated judgment systems.
Questioning Data Ownership With Art
Christopher Meerdo’s 2025 installation, Exuvium (blue), merges generative AI, physical sculpture, and offline data architectures to interrogate digital surveillance, technology control, and data rights.
Visitors encountering Exuvium (blue) observe glowing, suspended sculptural forms and discover four localized Wi-Fi access points on their mobile devices. Connecting to these networks unveils a browser portal loaded with public-interest files, including the 269-gigabyte BlueLeaks archive alongside thousands of Los Angeles Police Department internal chat logs and officer headshots.
There are no external internet connections, no cloud servers, and no corporate trackers monitoring their activity. I hope viewers leave understanding that data networks do not have to exist as extractive corporate infrastructure; they can be localized, physical, low-power, and community-controlled.
Meerdo explains that the installation bridges tactile physical sculpture with invisible digital data spaces, functioning ultimately as a counter-forensic intervention that reframes suppressed state records as public geocache nodes.
Investigating machine-learning architectures since 2018, Meerdo's archival fascination began with his grandfather's World War II negatives and a 1936 Rolleiflex camera. By 2024, he adapted this methodology by training a custom Low-Rank Adaptation (LoRA) model entirely offline on his home computer using his personal portfolio, keeping it completely isolated from corporate cloud infrastructure.
I think all of the artists working with building and running their own locally run AI models trained on their own data are effectively prototyping alternatives to the vision of AGI being sold to us by Big Tech.
By studying the visual glitches and bleeding edges produced by diffusion models, Meerdo translated these digital anomalies into hand-bent galvanized wire, fabric, and translucent polyethylene. This approach represents a structural transformation in the creative workflow rather than basic AI assistance.
Ultimately, Meerdo wants audiences to realize that artificial intelligence is not restricted to massive corporate cloud datacenters, emphasizing that it consists of modular models that can be reclaimed and deployed locally.