AI Images are Becoming Increasingly Difficult for Computers to Identify

There is a dreadful sense that the refrain "Didn't happen if there weren't pics" is about to become obsolete.

‍

The Wall Street Journal reported that even the programs designed specifically to identify phony images created by AI, which are developed for the purpose of identifying phony images, are unable to distinguish between real images and fake images as the world's largest tech companies continue to explore the capabilities of artificial intelligence. Currently, governments are taking steps to curb the future of technology.

‍

How do we define reality?

There are still a lot of free AI programs out there that produce pretty janky images. When you go to the artificial intelligence image generator site Craiyon, which was formerly known as Dall-E Mini, you might be tempted to enter in the name "Tony Soprano," and you'll probably get something that looks like the app mashed up 10 different photographs of the fictional mafia dad together with the smudge tool and then smudged them together.

‍

Founded in 1936, Artificial Intelligence has radically changed the way we think about the future. What was once the stuff of sci-fi writers like Arthur C. Clarke and Harlon Ellison has fast become routine. And like those writers, AI's potential for good as well as harm is becoming increasingly obvious. Despite fake images of the "Puffer Coat Pope," Donald Trump being tackled by police, and Emmanuel Macron running from protesters, it will still be difficult to keep up with the advances in artificial intelligence due to the fact that Microsoft, OpenAI, Alibaba, and others are investing a lot in the field.

‍

• As part of a project called Optic's Ai or Not, the company has a website which has an accuracy rate of 95%, but since the latest update by Midjourney's technology, it has dropped down to 89%. At one point, the website was even fooled by "Puffer Coat Pope" during one test round.

‍

• As a way to stay ahead of technology, companies like Microsoft impose restrictions on the generators on their sites, while Bing's Image Creator does not allow users to use prompts with well-known figures as prompts. The founder of Midjourney, David Holz, told the Wall Street Journal that the company is rolling out an algorithm to handle user requests that will replace the company's human moderators.

‍

Among the other companies that tag AI content is Hive, whose CEO Kevin Guo told the Wall Street Journal that the AI environment is a race between the tools. "We follow all the tools out there, and we have to update our models every time they update theirs," said Goo.

‍

Taking the Clampdown seriously: China adheres to its own guidelines with good faith, but it won't take any chances. A draft set of rules was released by the government of the Middle Kingdom on Tuesday that would require companies launching artificial intelligence tools to submit security assessments to the authorities before launching them into the public domain. A user would also have to provide his or her real name and other personal information before he or she could access any of the AI programs. In the event that companies don't comply with the regulations, they may be fined, suspended, or even investigated by law enforcement.