MUMBAI: Moviemaking may once have begun with a camera, but at IFFI this week, the consensus seemed clear: in the age of artificial intelligence, all you really need is curiosity and a prompt. At the 56th International Film Festival of India in Goa, the masterclass The New AI Cinema: A Discourse on Generative AI and LLMs quickly turned into one of the festival’s most electric sessions, a spirited, sometimes philosophical, often provocative debate about how AI is reshaping the craft, the industry and even the idea of storytelling itself.

Helmed by filmmaker V. Muralitharan, auteur Shekhar Kapur, and writer-director Shankar Ramakrishnan, the session moved far beyond technical demos. Instead, it cracked open the cultural, artistic and ethical tensions simmering beneath cinema’s AI revolution.

Muralitharan opened with a glimpse into AI’s unexpected role in education. He described sitting with an academic panel studying the human nervous system — a complicated animation traditionally achievable only with expensive modelling tools, now fully AI-generated. Neural networks visualising actual neural networks: the irony wasn’t lost on anyone.

He argued that the same technology could resurrect history, lost characters, forgotten struggles, major events and accidents, all recreated with precision as long as research materials were robust. “Our project is only just taking shape,” he said, hinting at a future where historical cinema no longer depends on scarce archives.

Muralitharan also offered a deeply personal confession. After a decade in corporate managerial roles, his hands-on artistic skills had faded. He felt the gap acutely. But AI, he said, had restored his ability to “make my films and tell my stories” without needing to relearn complex software.

Advertisement

Initially guided by engineers, he soon realised that modern tools “are like talking to a person”. Simple English prompts could now generate shots, scenes or entire sequences. “There is no technical expertise needed today,” he said. Creativity, not coding, had become the entry barrier and even that barrier was getting lower.

Kapur agreed. “It’s become very democratic,” he added, before launching into an unexpected tale of how AI fed his obsession with the relationship between physics, poetry and schizophrenia.

Kapur recounted spending three months on ChatGPT exploring the idea that Van Gogh, declared schizophrenic, may have intuitively painted the true nature of fluid dynamics — something physicists still struggle to mathematically resolve.

If art can reveal truths that science can’t express, Kapur argued, then AI becomes a tool not for shortcuts but for deeper inquiry. “I found relationships I had never learnt before,” he said, describing the chatbot as an engine of personal education.

While some debated continuity flaws in AI-generated shots like three dogs inexplicably shrinking to one, Muralitharan insisted the tools evolve weekly. Literally. “What I saw two weeks back is not the tool today,” he said, describing how character consistency across ages had recently become achievable.

Advertisement

Their seven-minute demonstration film featured a protagonist across three life stages, generated consistently something impossible even months earlier.

To create long continuous sequences, they generated eight-second blocks, each feeding the next. It took three or four cycles to create a smooth 20-second time-lapse with crowds, movement and controlled camera speed. Soon, he said, tools will generate arbitrarily long shots, though “they must be meaningful”.

One of the biggest misconceptions in public discourse, the speakers stressed, is confusing AI with VFX.

“AI is far, far more sophisticated,” Muralitharan said. VFX still requires cameras, green screens and physical shoots. AI requires nothing but data, imagination and prompts.

Yet the two are now merging. Most VFX pipelines already secretly incorporate AI for set design, costumes and environmental generation. Kapur admitted he had recently seen a commercial where even he couldn’t distinguish the real Hrithik Roshan from his AI double.

Advertisement

Kapur then pulled back the curtain on why OTT platforms and theatrical distributors wield such power: the barrier to entry is high because budgets are high. AI, he said, will bulldoze that barrier.

“A. R. Rahman and I are setting up an AI film school in Dharavi,” he announced, explaining that kids with no equipment, no training and no industry access will soon make films purely through AI tools.

“When budgets crash, gatekeepers lose control,” Kapur said. OTT platforms will acquire AI-made films, “they won’t have a choice.”

When asked whether AI can replicate great performances, Kapur answered with characteristic bluntness: AI can replace a superhero suit but not Shabana Azmi’s eyes.

“AI eyes have a dead look,” he explained, offering a scientific nugget: a human pupil performs around 1,000 micro-movements per minute movements so subtle even scientists rarely acknowledge them. AI, he said, cannot recreate that expressiveness anytime soon.

Advertisement

Close-ups of great actors will remain the human domain at least for now.

On intellectual property, Muralitharan emphasised responsibility. If a filmmaker wants to use someone’s image, they must seek consent and sign contracts. AI may be a mirror and a provocateur, he said, but it doesn’t erase ethics.

Kapur added that artistic inspiration has always been a form of borrowing. “We study every painter in art school,” he said. “Of course we’re influenced.” The problem isn’t influence, it’s inertia. “How much you depend on AI depends on how lazy you are,” he added, earning laughs.

Kapur warned that AI will transform how organisations, hierarchies and power structures function. Traditional pyramids thrive on inertia and AI “sucks inertia out of systems”. As roles blur, “you won’t know who leads and who follows anymore”. The shift, he said, will redefine how we live, work and learn.

One of the conversation’s most poetic turns came when Muralitharan spoke about his fascination with lost Indian cinema. Many early films, including the 1928 silent Tamil film Vigadha Kumaran, survive only in photographs. AI, he said, could reconstruct these films not perfectly, but evocatively enough to restore what history has forgotten.

Advertisement

Kapur acknowledged AI’s early Western bias but said India is quickly training its own sub-models. Today, he said, AI-generated characters increasingly “look very Indian”, and models are learning Indian language structures. The days of a tiger with white stripes, a famously wrong early AI output are fading.

The speakers did not shy away from AI’s energy demands. Kapur cited data showing AI prompts consume up to 10 times the energy of a Google search. Even small prompt choices matter. “Remove the word please,” he said. “It saves power.”

But he pointed to emerging quantum chip research, where chips compute “maybe” rather than binary “yes/no”, cutting heat and energy drastically. “It will change everything,” he predicted.

In a moment of déjà vu, a crew member reminded Kapur that 20 years ago he had predicted that digital filmmaking would enter every home. “It happened,” Kapur said. And now AI is doing it again but faster, deeper, and in ways that will reshape not just cinema, but society itself.

Advertisement