Ben Affleck has recently given several interviews in which he explained his hands‑on interest in artificial intelligence and how it connects to film production. Earlier this year he sold an AI filmmaking startup to Netflix for a reported $587 million; Affleck has said that figure is "not right" and noted he did not own the company outright.
Where he spoke and when
Affleck described his work on GQ’s “One More Question” series and in an interview at the Screentime 2026 conference in Los Angeles with Lucas Shaw of Bloomberg. Short clips from those conversations have gone viral, showing him using technical terms such as machine learning, neural networks, transformers, tensors, GPUs, inference and discussing model fine‑tuning.
Concrete steps he took
He says the migration from analog to digital filmmaking sparked his interest. Affleck explained he can write modest Python scripts and walked through how an image becomes a tensor — batch number, frame number, and each pixel’s red, green and blue values — the numerical inputs networks operate on. He contrasted convolutional neural networks, which he called precursors to transformers, and described how networks perform pattern recognition, edge detection and feature extraction (for example, identifying a window ledge to make green‑screen replacement easier).
Affleck also said that at times he used his Hollywood position to get early looks at emerging AI technology, including visits to OpenAI. Those encounters pushed him toward more direct involvement in the field.
His startup and ethical approach
According to Affleck, he and collaborators decided to build their own dataset to work in an "ethical" way that respected artists’ likeness and existing relationships. He said he raised financing and spent roughly eight months shooting with many cameras and equipment to create a dataset intended to serve as late‑stage training for open models, aligning training and inference code around specific tasks that would generate value.
At Screentime and in other clips he described how they fine‑tuned open video models by "unfreezing" the model weights and training only the final cinematic layer so the result met production standards. This approach, he said, lets filmmakers retain proprietary work while gaining the benefits of training a bespoke model for a given film.
Affleck reported using these techniques in post production on his film Animals.
What he worries about — and what he does not
Affleck framed his concerns in practical terms: he is worried about his children in school, the increase in A grades given in colleges (he cited a 30 percent rise over three years), learned helplessness, and responsible use. He explicitly dismissed apocalyptic scenarios such as Skynet and said he does not believe AI will meaningfully replace filmmaking; rather, he expects it to be additive.
Why this matters
Affleck’s combination of public profile, hands‑on technical interest and entrepreneurial activity is notable because few high‑profile actors so explicitly describe the technical details of machine learning applied to filmmaking. The viral clips drew attention because his technical vocabulary and accounts of dataset creation, model tuning and production usage contrast with more sensational public narratives about AI. His stated aim is to integrate AI in ways that serve filmmakers and preserve artistic standards.
