Gabe Greenberg (@gabegreenberg) on X

Ben Affleck walks through his actual technical knowledge of machine learning and computer vision in this interview. He's not just name-dropping — he explains how convolutional neural networks work for video effects, breaking down the concrete steps: converting images into tensors (numerical representations with batch, frame, and RGB pixel values), then running edge detection and feature extraction to identify objects like window ledges so they can be replaced during green-screen work. This isn't theoretical for him. He ran a visual-effects company before Artists Equity and spent years working with GPUs, so he picked this stuff up through hands-on experience as film transitioned from analog to digital. What's interesting is how he leverages his position. He mentions that Google and OpenAI are doing significant work with transformers for video generation, and he's used his celebrity to get private access to see what they're building. When he calls and says "Hey, it's Ben Affleck. Can I come see what you're doing?" sometimes people actually say yes. He's not pretending to be an expert — he admits his Python skills are "pretty shitty" — but he knows enough to have real conversations with people building cutting-edge AI video models. The broader picture here is that Affleck has genuinely engaged with the technical side of visual effects and AI for years, not just as a consumer but as someone who worked in the industry. His understanding of how tensors and CNNs work comes from practical application in filmmaking, which gives him credibility when he talks about where the technology is heading and why it matters for his work.