Recognizing AI Generated Content
Image #1: Elf Girl
Prompt: Generate an image of a woman with pale skin and red hair looking to the right. It should be only her shoulders and head. She should have white mascara, soft red lipstick, elf ears, and jewelry including a cross earring.
Answer
AI is on the left. Notice that there are issues in the fine details of the image, like the eyelashes and chain on the jewelry. There is also a uniform amount of detail across the whole image.
Image #2: Strawberry Bowl
Prompt: Generate an image of a bowl of strawberries. There should be two strawberries in the foreground and a red cloth in the background. The bowl is made of pale green ceramic.
Answer
AI is on the left. Notice that the leaves on the strawberries blend together. The color of the strawberries is also very uniform and they look glossy with little texture.
Image #3: Antelope Canyon
Prompt: Generate an image of a rock formation like antelope canyon.
Answer
AI is on the right. There is a lot of repetition in the canyon shapes and there is little fine detail in the rocks upon zooming in.
Image #4: Malaysian Skyline
Prompt: Generate an image of the Kuala Lumpur skyline with the focus on the Petronas Twin Towers. The image should be taken at night, with a dark blue sky and the buildings lit up.
Answer
AI is on the left. Text is garbled, there is unexpected symmetry in the buildings, and upon zooming in the roads look like swirls.
Image #5: Arctic Fox
Prompt: Generate an image of an arctic fox sitting in the snow, looking at the camera.
Answer
AI is on the right. The fox has a yellow tinge to it, which many AI images do (colloquially known as the ‘piss filter.’ The fox is very symmetrical, and the toes are slightly garbled.
