Artificial intelligence (AI) has been increasingly good at fooling people. A series of photos showing former US president Donald Trump being aggressively arrested by police have caught people’s attention. They were fake but very convincing.
Created by the AI program Midjourney, the photos were highly realistic, from the characters’ movements to the surroundings. However, many details can give away the fact that they are made by AI. The Washington Post’s technology writer Shira Ovide shared her tips. The main idea is to spot the problems-anything that would look strange in a photo.
AI software has a history of generating human hands incorrectly. It sometimes can create hands with more than five fingers. This is because AI isn’t sure what a “hand” exactly is, according to Popular Science. The data AI uses to learn often show hands and fingers in various gestures, which can be very confusing for AI.
AI-generated images also usually contain details that are against reality. To spot this, focus on items like accessories. For example, people in an image may be missing earrings or one part of their sunglasses. If there’s text in an image, such as a newspaper or poster, it’s usually garbled (篡改), even though the text may look realistic from a distance.
Another thing AI is terrible at handling is the background. If there’s a crowd in the image, people’s faces in the background are usually blurry — or they don’t have faces at all!
The development of AI-generated art also raises alarm bells about how these fake images could be used to spread misinformation. “I think misinformation is going to hit an all-time high,” Jamie Cohen, a digital culture and AI expert in the US, told New York Post. Generating an AI artwork is to “create reality”, Cohen argued, adding that, being able to tell whether the work is real or not requires high media literacy (素养) skills. “The world may not be ready for how realistic the images have become,” Shane Kittelson, a US researcher, told The Washington Post.
8. Why are Donald Trump’s photos mentioned in the first paragraph?
A.To draw more people’s attention. |
B.To make people convinced of the news. |
C.To show the AI-generated images were fake. |
D.To prove that it’s easy for AI to cheat people. |
9. Why does AI generate human hands incorrectly according to the text?
A.It is unable to recognize human hands. |
B.The data it uses to learn contain errors. |
C.It doesn’t fully understand human hands. |
D.It has insufficient data about human hands. |
10. Which of the following will Jamie Cohen agree with?
A.AI artworks may not be reflective of true creativity. |
B.It’s high time to address the risks posed by AI works. |
C.AI artworks will replace traditional art forms in the future. |
D.Detecting real art demands strong media literacy competence. |
11. What is mainly talked about in the text?
A.AI is tricking people with highly realistic photos. |
B.AI-generated art make people worried about fake images. |
C.AI is terrible at handling the background. |
D.The world may not be ready for AI. |