Can you guess which of these paintings was not made by a human?

Can you tell the difference?
Can you tell the difference?
Image: GumGum
We may earn a commission from links on this page.

Take a look at the paintings above. Five were painted by artists from around the world. One was painted by a robot. If you can’t tell the difference, you’re not alone.

The pieces were commissioned as part of an experiment by GumGum, a Santa Monica-based artificial intelligence company, and the Art & Artificial Intelligence Laboratory at Rutgers University. The researchers were not attempting to answer whether a machine could approximate a human’s final product, but rather whether a machine could approximate a human’s creative process.

The process, after all, is what makes art uniquely human. For us, art is not an end in itself; it’s a means to communicate a shared experience.

“Machines don’t know what it feels like to experience the death of a newborn or wait for a new album to come out,” says Ben Plomion, chief marketing officer at GumGum. “Because of that, I don’t think machine-generated art can ever be as richly appreciated as something made by a human. But machines can certainly help humans in the process of creation.”

All six artists participating in the experiment were commissioned to paint a piece inspired by the same collection of 20th-century American abstract expressionists. For Cloudpainter, a painting robot developed by Virginia-based artist Pindar van Arman, the collection became a dataset to train its algorithm. Its final output (painting F above) is a far cry from the geometric, color-between-the-lines art you might imagine from a robot artist. Instead, with dripping colors and blurred lines, the piece looks surprisingly, well, human.

For Cloudpainter, as with many automated technologies today, there is still a human in the loop. Van Arman built the robot and is responsible for programming the software that dictates its output. As a human-robot collaboration, Cloudpainter is a fairly accurate reflection of what automation looks like today.

Rather than the human-work vs robot-work binary often perpetuated by doomsdayers’ fear of the robocalypse, examples of modern-day automation tend to look more like cyborgs—with humans and machines working together to accomplish more than either could do individually. Whether the robot is a pizza chef or a chess master, machines can augment human work rather than just replace it.

When it comes to art, Plomion imagines robots can play the role of virtual assistants—providing inspiration for the artist and performing menials tasks like shading in a background or mixing a specific color. He believes the definition of human creativity will evolve with the technology. There will be new creative decisions to be made about how to artfully use these new tools.

Regardless of how it makes you feel, robot art is already a reality. Last Month, Christie’s auctioned off a piece of AI-generated art for the first time. (It sold for around $10,000.) But the question remains whether we’ll reach a day when machine-generated art needs a different name.

“It’s important to realize how much intention plays into our understanding of art,” says Dylan Freedman, a programmer who researched AI-generated art at Google. “Truly machine-generated art without any human intent behind-the-scenes doesn’t exist yet. When it does, we’ll have to grapple with whether it’s still art at all.”