Nature issues AIGC ban! Submissions that use AI for visual content will not be accepted

As one of the most authoritative scientific journals, Nature recently made it clear:

prohibits the use of image and video content created by generative artificial intelligence (AIGC)!

This also means that, except for articles whose topics discuss AI, any work accepted by Nature must ensure that there is no visual content generated or enhanced by AIGC.

This objection vote is filled with a line of big words:

Integrity, permission, privacy and intellectual property protection

Some netizens believe:

In fact, this is a new step taken by us to re-discuss the "truth claim" of photography.

Digital photography and Photoshop both changed our relationship with media, and AIGC may do it again.

The popularity of ChatGPT has promoted AIGC’s strong “out of the circle”, and all walks of life are competing to explore its potential. Controversy has persisted over whether AIGC should be used to display visual content in science, art, publishing, and other fields.

Concerns about AIGC are inseparable from the intensification of infringement caused by irregular use of AIGC in the past six months.

As early as January this year, in the Northern District Court of California, three cartoonists launched a class action lawsuit against three AIGC commercial application companies, including Stability AI, accusing Stable Diffusion of infringement. Similar cases are common.

Last month, Douyin, one of the most popular short video social platforms at the moment, proposed eleven platform specifications and industry initiatives:

Become a creator and anchor , users, merchants, advertisers and other platform participants must comply with 11 specifications when using the generative artificial intelligence technology applied by Douyin.

This includes prominently marking the content generated by AIGC, and prohibiting the use of AIGC technology to create and publish infringing content, as well as content that violates scientific common sense, practices fraud, and spreads rumors.

Nature, which has a history of 153 years, has only a simple "no" towards AIGC.

NatureWhy should we disable AIGC?

After generative AI tools such as ChatGPT and Midjourney have been widely used and their capabilities have grown rapidly, Nature cannot sit still.

There have been months of intense discussions and consultations on this issue.

The final result is:

Unless it is an article specifically about artificial intelligence, Nature will not be able to write articles specifically about artificial intelligence in the foreseeable future. No photography, video or illustration content created in whole or in part using AIGC will be published.

Nature will not allow the use of AIGC in visual content, and the reason boils down to:

Integrity issues.

Whether scientific or artistic creation, the publishing process should be based on a shared commitment to integrity.

One of them is to keep the process transparent.

Ensuring the accuracy and authenticity of data and image sources is the shared responsibility of researchers, editors, and publishers. This is something that existing AIGC tools cannot do.

The AIGC tool does not provide a way to access data and image sources, so this verification is not possible.

Another big problem is the issue of attribution.

The importance of accurate attribution of sources when using or citing existing works cannot be ignored and is a core principle in scientific and artistic publishing.

Obviously, the content generated by the AIGC tool cannot clearly identify the ownership issue.

Consent and permission is also one of the factors that must be considered.

If intellectual property content is involved, consent and permission must be obtained, and AIGC once again failed to meet expectations on this issue.

The AIGC system is trained on images from unidentified sources. AIGC often uses some copyrighted works for training without permission. There are situations where privacy rights may be violated, such as using someone else's photos or videos without their permission.

In addition to privacy issues, these "deep fake" contents can also easily accelerate the spread of false information.

One More Thing

Although Nature does not accept visual content generated by AIGC, it does allow the text to contain content assisted by AIGC.

Provided that some caveats are followed:

The use of such large language model (LLM) tools needs to be documented in the research methods or acknowledgments section of the paper, and we expect authors to provide all Sources of data, including data generated by assisted artificial intelligence. Furthermore, any LLM tool will not be accepted as the author of the paper.

Nature believes that this is done to protect content creation results. The world is on the brink of an artificial intelligence revolution that holds great promise but is also rapidly upending long-established traditions in science, the arts, publishing, and more. Careless use of AI could unravel centuries of systems that protect scientific integrity and protect content creators from exploitation.

What do you think of it?

Reference link:
[1]https://www.nature.com/articles/d41586-023-01546-4
[2]https://arstechnica.com/information-technology/2023/06/nature-bans-ai-generated-art-from-its-153-year-old- science-journal/

The above is the detailed content of Nature issues AIGC ban! Submissions that use AI for visual content will not be accepted. For more information, please follow other related articles on the PHP Chinese website!