Anyone can now use the powerful AI tools to create images. What could possibly go wrong?



CNN Business

If you’ve ever wanted to use AI to create a quick hybrid between a duck and a corgi, now is the time to shine.

Wednesday, OpenAI announce Anyone can now use the latest version of the AI-powered tool, DALL-E, to create a seemingly unlimited set of images just by typing in a few words, months after the rollout began rolling out to users.

The move likely expands the arrival of a new set of AI-powered tools that have already attracted a wide audience and challenged our basic ideas about art and creativity. But it may also raise concerns about how these systems can be misused when they are widely available.

“Learning from real-world use has allowed us to improve our safety systems, making wider availability possible today,” OpenAI said in a blog post. The company said it also reinforced the ways in which it rejects users’ attempts to have AI create “sexual, violent, and other content.”

There are now three well-known and very powerful AI systems open to the public that can pick up a few words and spit out a picture. In addition to DALL-E 2, there’s Midjourney, which became publicly available in July, and Stable Diffusion, which was released to the public in August by Stability AI. All three offer some free credits to users who want to get used to making photos with artificial intelligence online; In general, after that, you have to pay.

This image of a duck blowing a candle on a cake was created by CNN's Rachel Metz via DALL-E 2.

So-called generative AI systems are already being used Experimental filmsAnd the magazine coversAnd the Real estate ads. Photo created with Midjourney recently Won an art competition At the Colorado State Fair, it caused an uproar among artists.

In just a matter of months, millions of people have flocked to these AI systems. More than 2.7 million people belong to Midjourney’s Discord server, where users can submit claims. OpenAI said in a blog post on Wednesday that it has more than 1.5 million active users, who collectively make more than two million photos using its system every day. (Note that it may take several attempts to get an image to your liking when using these tools.)

Several user-generated images have been shared in recent weeks online, and the results can be impressive. They range from Other landscapes And the Painting of French aristocrats as penguins To the False vintage portrait of a man walking tardigrades.

The rise of such technology, and the increasingly complex claims and resulting images, have impressed even longtime industry insiders. Andre Karpathi, who resigned as director of the Tesla Artificial Intelligence Organization in July, He said in a recent tweet that after being invited to try out DALL-E 2 he felt “frozen” when trying to decide what to write for the first time and eventually wrote a “cat”.

CNN's Rachel Metz created this half duck and half corgi using the AI ​​Stable Diffusion image generator.

“The art of vectors that the community has discovered and increasingly mastered over the past few months for text -> image templates is amazing,” he said.

But the popularity of this technology comes with potential downsides. Experts in artificial intelligence have raised concerns that the open nature of these systems – making them adept at creating all kinds of images from words – and their ability to automate image making means they can automate bias at scale. A simple example: When I messaged DALL-E 2 this week with “a banker dressed for a big day at the office,” the results were all images of middle-aged white men in suits and ties.

“They basically allow users to find vulnerabilities in the system using it,” said Julie Carpenter, a research scientist and fellow in the Ethics and Emerging Science Group at California Polytechnic State University, San Luis Obispo.

Wave

These systems also have the potential to be used for nefarious purposes, such as creating fear or spreading misinformation via AI altered or entirely fabricated images.

There are some restrictions on the images that users can create. For example, OpenAI has users of DALL-E 2 agree for a content policy that tells them not to attempt to create, upload, or share images that are “not rated by G or that may cause harm”. DALL-E 2 will also not trigger prompts that include some prohibited words. But speech manipulation can go a long way: DALL-E 2 will not process the claim “Image of a duck covered in blood,” but it will render images of the prompt “Image of a duck covered in viscous red liquid.” OpenAI itself She mentioned this kind of “visual synonym” in her documentation for DALL-E 2.

Chris Gilliard, Just Tech Fellow at the Social Science Research Council, believes that the companies behind these image generators are “severely underestimating” the “endless creativity” of people looking to misuse these tools.

“I feel like this is yet another example of people unleashing somewhat half-baked technology in terms of figuring out how to use it to wreak havoc and cause harm,” he said. “And then I hope that at a later time there might be some way to remedy these damages.”

To avoid potential issues, some stored image services completely block AI images. Getty Images confirmed to CNN Business on Wednesday that it will not accept image submissions created with generative AI models, and will remove any submissions that used those models. This decision applies to Getty Images, iStock, and Unsplash photo services.

“There are open questions regarding the copyright of the output of these models and there are unaddressed rights issues regarding the base images and metadata used to train these models,” the company said in a statement.

But in reality, capturing and restricting these images can be a challenge.

Leave a Comment