Ben Affleck's Secret AI Startup: Python, Tensors, and Open-Weights
Hollywood actor and director Ben Affleck reveals his deep technical expertise in AI, detailing how he learned Python, studied neural networks, and built a copyright-safe open-weights video AI startup.
Ben Affleck is not just a Hollywood A-lister; he is a hands-on tech enthusiast who writes Python code, studies neural networks, and has developed a proprietary AI video technology that respects copyright laws. In a recent interview with GQ magazine, the actor and director shared his deep dive into the world of artificial intelligence, tracing his journey from analog filmmaking to digital VFX workflows and machine learning.
From Visual Effects to AI
Affleck recounted how his fascination with computers led him to explore convolutional neural networks, tensors, and transformers. After consulting with industry insiders and tech giants, he realized the disruptive potential of generative video models, famously telling his producing partner Matt Damon that the traditional filmmaking landscape was about to change drastically. Despite the initial shock, Affleck identified critical gaps in how major tech companies approached cinematic data.
I realized this was incredible technology created by incredible people, but they lacked the specific domain expertise in filmmaking, Affleck noted.
Open-Weights and Clean Datasets
Refusing to train models on copyrighted works without permission, Affleck and his startup assembled a clean dataset over months of meticulous work, utilizing specialized cameras and LiDAR scanning technology. Speaking at a recent Bloomberg conference, Affleck explained that his startup utilized open-weight AI models, performing model unfreezing to strip away proprietary pre-training data and replace it with their custom cinematic dataset. This innovative pipeline was already put into practice in Affleck's latest Netflix feature, Animals.
