AI creates images in seconds – magical and dangerous

Avatar of Nick John By Nick John Dec22,2023 #creates #dangerous
AI creates images in seconds - magical and dangerous 0
AI creates images in seconds - magical and dangerous 0

Two photos created by Dall-E based on descriptions.

`Red and yellow bell peppers placed in a floral patterned bowl on a green rug`, or `A woman wearing a red jacket looking up at the sky standing in the middle of Times Square` are the descriptions for the two photos above.

Since OpenAI launched Dall-E 2 in April, the community has been excited, even startled, when this tool created a series of images with unbelievable realism.

Five months after Dall-E 2 was announced, there were 1.5 million users on the platform and creating more than two million photos a day.

The birth of Dall-E 2 also caused the trend of turning text into images to explode.

The magic of photos from AI

Abran Maldonado is an artist specializing in using AI to paint in New Jersey.

“Oh my God, this AI is going to get me fired,” the reporter exclaimed, followed by a laugh.

AI creates images in seconds - magical and dangerous

The four photos were created by Dall-E on Maldonado’s orders.

Dall-E created four versions of the image upon request.

Previously, a Google engineer evaluated the LaMDA AI chatbot as having `child-like awareness` in June, causing a major controversy about the risk of robots one day dominating the world.

Both deepfakes and programs that create images from text rely on an AI training method called deep learning.

Worry

New AIs with upgraded features every day are causing great concern.

`Once the line between real and fake is blurred, everything will become fake. We will not be able to believe anything in this life,` admitted professor Wael Abd-Almageed of the University of Southern California.

According to OpenAI representatives, the Dall-E tool is trying to push back those concerns.

`With AI, you have to let it learn from exposure to reality, through what users do with it and how it pushes its limits,` said Sam Altman, CEO of OpenAI.

However, other AIs are not like that.

Efforts to control

In February, OpenAI invited a `red flag group` of 25 researchers from industry leaders to examine Dall-E’s flaws.

Among the biggest concerns, the team found that Dall-E could be used to create targeted harassment, bullying, and exploitative content.

OpenAI then built-in filters, blocks, and flagging systems into its AI.

According to an announcement in June, OpenAI is confident in its ability to intervene if things do not go as expected.

Maldonado advocates limiting the use of real faces on Dall-E to prevent them from being exploited for malicious or misleading purposes.

Another famous photo-from-description app is Midjourney.

Related Post

Leave a Reply

Your email address will not be published. Required fields are marked *