As the above AI portrait paintings proved, GANs aren’t just for evil anymore. Now they’re designing chairs, too. In the ChAIr Project, furniture designers Phillip Schmitt and Steffen Weiss trained a generative adversarial network on 562 images of classic 20th-century chairs, then pitted the dueling algorithms to generate new designs and then verify and optimize them, respectively. The final results are hilariously odd and often hard to actually sit on, but surprisingly captivating in a 20th-century design mashup kind of way. So how to get these AI chair designs to MOMA? Training, training, training.
sign up for our newsletter