Pindar Van Arman has spent more than a decade teaching robots to become painters. He’s trying to build real artists — not glorified printers or Instagram filters but algorithms that make creative decisions and develop their own style over time.
Van Arman is a computer security contractor in northern Virginia. The hardware for his robot artists is pretty simple, and it can be built for a few hundred dollars. But it’s the software that’s complex. It can apply the style of one image — say, Van Gogh’s “Starry Night” — to another, like a photo of someone’s face. A camera takes photos after each brushstroke, so the AI can analyze its work as it progresses.