Albert Serra and Bi Gan Debate Literary Adaptation and Why AI Will Never Have Innocence
Get the Culture newsletter
Daily culture — film, music, books, the trends and ideas worth your attention. Free.
- Albert Serra and Bi Gan sat down together at the Shanghai International Film & TV Market for the panel "Stories Travel Further: Literature & Cinema in Spain-China Dialogue," having met for the first time just a month earlier in Paris.
- Serra dismissed literary adaptation as work "for lazy people because they don't want to figure out an original idea," saying he uses classic texts only as a departure point for something entirely new.
- Bi Gan took a "gentler line," treating literature as structural rather than reverential, often borrowing book titles as film titles and weaving poetic narrative structures into his work.
- Both directors converged on the counterintuitive claim that mediocre books produce better films than great ones, because authors of good books "feel inside a prison" and become too respectful of the source material.
- Serra declared "the only thing AI will never have is innocence, because AI is based on collecting data, and innocence is based upon deleting data," arguing real artists destroy what came before them.
- Bi questioned whether human-AI communication is even possible, calling the miscommunication "unsolvable" and a "natural contradiction" inherent in the process.
- Bi credited Federico García Lorca's poetry — described as "brief and beautiful, as a small, soft outcry" — as shaping his sensibility on mortality and fear, even when the influence isn't legible on the page.
Why it matters: The panel paired two of arthouse cinema's most distinctive voices — Serra, who treats canonical texts as raw material to betray, and Bi, who builds poetic structures from literary scaffolding — to argue adaptation works best when directors feel free to ignore their source. Their shared AI skepticism — innocence requires deleting data, not collecting it — frames originality as destruction, which automation cannot replicate.
Ask SkimNews
