GPT2Draw is a collaboration between artificial intelligence and human input. It consists of a dataset of drawings, drawn by humans whom were given instructions from a neural network (the gpt2) on how to draw different objects. The A.I tool generated the instruction text which people then drew from. The people had no conscious idea of what they were supposed to draw, resulting in a very divergent dataset of drawings.
The goal of these drawings is to explore why we rely on computers to help us make every day decisions: What to eat for breakfast? What to wear outside? How to draw body porportions? Neural networks like search engines can be a helpful tool, but when does it become more ignorant than helpful to trust in computerized strategy?