Experimenting with AI tools is one of my favorite parts of my practice, and this particular video generator turned out to be a very cool collaborator. It is not 100 percent accurate, but it gets surprisingly close, and that “almost right” quality ended up becoming part of the interest for me. NotebookLM is a Google product, available at https://notebooklm.google.com, and it works differently from a general chatbot like Gemini or ChatGPT, because it builds everything from the specific sources you feed it rather than from the entire internet.
NotebookLM’s video overview tool ended up acting like a surprise co‑director for a project where I wanted to explain proto Jewish futurism in the context of the Vitebsk People’s Art School. Instead of hand crafting every frame, I loaded it up with my recent article and asked it to propose a first pass at a lesson: a narrated video that walks viewers through how Jewish, revolutionary, and avant garde energies briefly converged in Vitebsk. To be more specific, it appears to be a combination of the slideshow generator + a podcast voice generator combined to make a “video”. The result was messy in places, visually strange, and full of small errors, but it still managed to deliver my main argument about Jewish futurist tendencies in this short and intense moment of art education.
Setting up the experiment
I started with a pretty simple goal: turn a dense, theory heavy pile of notes on Vitebsk, Marc Chagall, and the Russian avant garde into something a non specialist could actually watch and follow. NotebookLM’s promise of an auto generated video overview sounded like the right kind of constraint and collaborator for that task. Because it works on data that you explicitly upload or link, I gathered my materials into one place historical sources, exhibition texts, and fragments from my ongoing writing on Jewish futurism and treated the notebook as a compact archive that the system could mine for a narrative, instead of letting it improvise from generic web knowledge.
From notes to video overview
With the sources in place, I asked the system to create an explainer style video focused on three threads: the Vitebsk People’s Art School, the artists at its center, and the ways their work points toward possible Jewish futures. What came back was a sequence of slides paired with narration that moves through the post revolutionary context, the founding of the school, and its radical pedagogical experiments. What surprised me was how clearly the structure echoed my own framing of Vitebsk not as a footnote in art history, but as a kind of prototype for Jewish modernity in motion.
AI generated imagery close, not exact
The visuals were where the experiment really got interesting. The system did not reach for actual archival photos or specific paintings. Instead, it produced images that felt like approximations of the artists’ styles. Scenes appeared that looked almost like Chagall’s floating shtetl figures, nearly like Lissitzky’s architectonic compositions, and somewhat like Malevich’s abstractions, but never fully matched the originals. That almost quality created a kind of productive uncanniness. The video builds an atmosphere of Vitebsk’s avant garde world without literally reproducing it, more like a synthetic memory or dream constructed from stylistic cues, which for a project about Jewish futurism feels conceptually on point.
Glitches, spelling errors, and the shape of the argument
The video is clearly not a polished museum product. There are spelling mistakes, clunky phrasing, and the occasional slightly wrong name or term. For me, that did not invalidate the experiment, it just made the mediation visible. This is a generated draft that still needs a human editor, not an authoritative final cut. What mattered more was that underneath those glitches, the bones of the explainer were solid. The video successfully communicated the ideas I wanted to surface: the social and political context of the school, the specifically Jewish dimension of the work, and Vitebsk as a site of radical possibility rather than a nostalgic lost world.
Surfacing a Jewish futurist reading of Vitebsk
The real test for me was whether the piece could carry my reading of Vitebsk as a proto Jewish futurist project. The video condensed that framing into clear, accessible language, repeatedly returning to the school as a laboratory for new Jewish forms and new ways of being together. By forcing my notes into a short, watchable format, the tool pushed me to concretize what I actually mean by proto Jewish futurism in this context: an art school that treats Jewish life as material for design, not just content for preservation, and that treats pedagogy itself as a kind of speculative world building.
In the end, the experiment showed me how an AI generated video can function both as an explainer and as a mirror for my own thinking. It reflected my argument back in another medium, making it obvious which ideas translated smoothly into narrative and which still need more nuance and friction. The off brand imagery, the typos, and the overall coherence all became part of the story, a contemporary, imperfect, and strangely fitting echo of the Vitebsk school’s own attempt to invent a new way of seeing Jewish futures.
If you haven’t tried NotebookLM, I’d make an account and try some experiments on your own.
