Nvidia's new AI creates incredibly fake fake videos

Nvidia is a powerful company in the field of graphics hardware, possessing extremely intelligent AI, able to create images of people who never existed.

Recently, once again, the online community was taken aback by easily changing the weather scene, turning night and night into videos, and even changing spots on a leopard!

Nvidia's AI easily changes the weather scene, turning day and night.

According to TheNextWeb, researchers at Nvidia have developed an "unsupervised" method for computers, giving it the ability to change the content of videos it has learned. By this new method of learning, Nvidia's AI has produced incredibly realistic fake videos:

With the previous conventional methods, AI will need huge amounts of data by comparison. Moreover, computer training finds patterns - the basis of unsupervised learning - is also very difficult.

According to researchers, this problem is similar to teaching a machine how to color a picture:

"For example, super-resolution can be considered a problem of mapping a low-resolution image to a corresponding high-resolution image; coloring can be considered a problem of map a gray-scale monochrome image to a corresponding color image ".

Currently, Nvidia computers have been able to turn sunny days into rainy days, or create filters to create extremely impressive snowfall scenes, turn original video clips into new videos, though fake but real! Imagine what a machine can do in a decade?

There is a fact that worldwide, people are training computer tricks to deceive the human eye. Nvidia's new achievement in the field of artificial intelligence has shown a leap, giving AI the ability to reproduce reality to the startling level.

Oli Franklin-Wallis, of Wired, said: "The biggest damage the AI ​​has caused is not losing your job, but completely erasing your trust in whatever you hear. see "!

People are training computer tricks to deceive the human eye.

Perhaps the machine has not reached the level of "Matrix" movie, but over time, who knows?