Top AI News, December 2022 

Welcome to the overview of what the world of AI has been up to this month.

Adobe Stock Accepted AI-generated Content on Its Platform

As stated in a blog post, Adobe Stock is willing to accept content created using generative AI tools as long as it meets the company’s submission standards. 

To submit AI content, creators must have the necessary rights to license it for commercial use and be aware of the terms and conditions of the generative AI tools they worked with. They must also label the content as being created with AI and identify it as an illustration, rather than a photograph — even if it looks like a photograph. Images that depict or are based on an identifiable person must have a model release, while pictures depicting a fictitious person must be submitted with a property release. Adobe strictly forbids submitting AI-generated content for editorial use — to the Illustrative Editorial Collection (IEC) designed to illustrate articles on current events and newsworthy topics. 

Adobe Stock is not a pioneer among stock image giants who started adopting AI content. A few months earlier Shutterstock announced its partnership with the San-Francisco-based organization OpenAI. According to Shutterstock’s press release, the company will integrate the AI image generation capabilities of DALL-E (a text-to-image model built by OpenAI) in the coming months. Simultaneously, another global visual content agency, Getty Images, partnered with BRIA, a developer of creative AI tools. BRIA and Getty Images aim to develop tools that harness the power of AI and democratize the creative process. 

OpenAI Released a 3D Model Generation Tool

Machine learning research and development lab OpenAI has been making waves in press and social media earlier this year after delivering the text-2-image model DALL-E and the natural language process model ChatGPT. Now it gave us a new tool that might disrupt AI content creation field. A new algorithm Point-E is capable of creating a 3D representation of objects from a text prompt — you can play around with the lightened version of the algorithm on HuggingFace.


Point-E creators have paired a text-to-image model with an image-to-3D model. The first one generates an image from a text prompt, which is then transformed into a point cloud using the second model.

OpenAI wasn’t the first company that built a text-to-3D model, but what’s groundbreaking in their research is operational efficiency. 3D models can be produced in a couple of minutes on a single GPU, whereas with the Stable DreamFusion algorithm powered by Stable Diffusion, it takes about 3 hours to produce a 3D model.  

The Backlash of the Art Community Against AI Content

ArtStation, a portfolio platform for digital artists, faced a protest against AI-generated artworks. As ArtStation’s content guidelines don’t prohibit the use of AI, this kind of content has repeatedly reached the top of the platform’s main page which made human artists outraged.

Users have been spamming their portfolios with the no AI symbol to draw attention to the issue.

The ArtStation’s main page, December 19, 2022
Some artists came up with extraordinary ideas such as images poking fun at the fact that AI can’t draw fingers. Source: Veronika Kozlova, ArtStation

ArtStation responded to the protest by posting a statement where they stood for their principle not to forbid the use of AI content. Although, they insisted users need to have permission to publish their work created with AI. For those who don’t want AI companies to scrape their content, the platform suggested adding a “noAI” tag to their projects. Apparently, some artists remain dissatisfied with the response and want the company to take a stronger stance against AI-generated artwork by banning it. 

As of today, just a few protest pieces on ArtStation’s main page are shown. Nicholas Kole, one of those artists who sparked the protest, noted that “it’s been moderated back to silence”.

ArtStation wasn’t alone in being criticized by the art community this month. There was a lot of controversy around the content used in datasets to train AI models. For instance, people pointed out that portraits created with the popular Lensa AI app include original artists’ signatures. There is complete uncertainty as to whether the artists were compensated for taking their work to train algorithms.

Q&A Website Stack Overflow Temporarily Banned AI-generated Answers

Stack Overflow, a platform where users help each other to find solutions to programming problems, paused accepting content created with the most recently released ChatGPT, as stated on the FAQ page. The volume of AI-generated answers on the website reached thousands. However, the content itself often lacks accuracy and doesn’t meet the platform’s standards, though delivered cogently. Being flooded with the influx of answers, moderators couldn’t carefully examine and verify their correctness. 

“The fact that the answers often require a detailed read by someone with at least some subject matter expertise in order to determine that the answer is actually bad has effectively swamped our volunteer-based quality curation infrastructure”, noted in the statement. 

Meanwhile, Stack Overflow acknowledges that the policy regarding the use of AI text generators may be changed after discussing it with the community. Until then moderators are empowered to issue immediate suspensions of up to 30 days to users who are bringing AI content onto the website. 

Meta Devotes 20% of Its Budget to Metaverse, While “VR Pioneer” John Carmack Is Leaving Meta

Source: Unsplash

Despite all the skepticism about the metaverse, Meta keeps investing in its vision of the concept. The company shows huge financial losses on Reality Labs (more than $9 billion over the first nine months of 2022). According to Meta’s blog post, in 2023, 20% of Meta’s overall investments go toward Reality Lab which focuses on building metaverse-related projects in the fields of virtual reality, augmented reality, and social platforms.

Meanwhile, John Carmack known as a “VR pioneer” is leaving Meta after nearly a decade working on VR products. He originally worked as CTO at Oculus, a company developing VR headsets. It was acquired by Meta (formerly Facebook) in 2014 with Carmack getting pulled into it. In 2019, he stepped down from his role as the CTO of Oculus to shift his focus to developing artificial general intelligence and moved to a consulting CTO position with Oculus. He has been repeatedly criticizing Meta’s management. His internal post announcing his resignation reflected a state of tension between him and his employer: “It has been a struggle for me. I have a voice at the highest levels here, so it feels like I should be able to move things, but I’m evidently not persuasive enough”. After leaving Meta, Carmack will focus on his own startup Keen Technologies which has raised $20 million

Spread the word