Talk: Unreliable Past

Lev Manovich explores generative AI through image series that connect personal memory, historical rupture and the limitations of statistical models. He relates his return from cultural analytics to individual image making to the shock of Russia’s invasion of Ukraine and a renewed sense of losing his former homeland. Rather than treating AI as a passive tool, he describes a struggle with systems that reproduce frequent visual patterns and substitute familiar clichés for unusual intentions. Ambiguous prompts, multiple reference images and deliberate engagement with errors help him resist this tendency. Images of damaged manuscripts evoke unrealised cultural projects, while devastated Russian landscapes turn stereotypical churches into an unsettling response to war. A garden series transforms instructions about fine lines into worlds literally constructed from lines, exploiting repetition and nested frames produced by zooming. Discussion links these miniature environments to architectural training, migration and psychological fragmentation. Manovich distinguishes simulated marks from material artworks that register embodied gestures, and values earlier AI models’ unpredictable misunderstandings. He also questions artistic pluralism: when every style is permissible and image generation offers limitless possibilities, choosing what matters becomes difficult. Personal experience provides a possible constraint, grounding work in concerns that can acquire significance beyond individual biography.

AI Generated Summary