Orkhan Mammadov, an Azerbaijan-born new media artist now showing “Visions” in Art Club Houston’s mirrored infinity room, works in a territory a lot of AI-and-tradition art projects gesture at but rarely fully commit to: training generative systems on centuries-old visual craft, then feeding the results back into that craft’s living practice rather than treating the tradition as raw material for a purely digital output.
What “Visions” actually does
Described as merging a thousand years of art history with machine learning, “Visions” continues Mammadov’s broader practice of transforming collective cultural memory into immersive aesthetic experience — using the infinity room’s mirrored architecture to multiply and extend generated imagery into a space that visually reads as boundless, rather than a fixed screen or frame.
Why Carpetdiem is the clearest version of his method
Mammadov’s earlier project Carpetdiem makes the two-way exchange explicit: his algorithms learn patterning from the ancient tradition of carpetmaking, while the tradition itself gains new patterns through AI-designed templates — artisans trained in the traditional craft then physically weave what the algorithm generates. That’s a meaningfully different relationship than AI art that samples a craft’s visual style and outputs a digital image referencing it; the actual deliverable loops back into physical, hand-woven objects made by the people who carry the tradition, extending what those weavers have practiced over millennia rather than replacing or merely quoting it.
Why this two-way loop matters more than a one-way homage
A lot of generative art that engages with cultural or craft traditions treats them as a style to be learned and reproduced — the tradition supplies the aesthetic, the algorithm does the generating, and the output stays entirely in the digital or gallery space. Mammadov’s model routes the algorithm’s output back through the same human hands and craft knowledge that trained it in the first place, which keeps the tradition an active, still-evolving practice rather than a frozen dataset. That distinction — treating artisans as ongoing collaborators rather than a historical source to extract from — is a genuinely harder version of “AI meets tradition” to execute, and a more honest one.