ChinAI #200: Will AI Kill Visual China Group?
Plus, celebrating 200 issues of ChinAI!
Greetings from a world where…
First forget inspiration. Habit is more dependable. Habit will sustain you whether you're inspired or not. — Octavia E. Butler.
Onward to the next 100 issues!
…As always, the searchable archive of all past issues is here. Please please subscribe here to support ChinAI under a Guardian/Wikipedia-style tipping model (everyone gets the same content but those who can pay support access for all AND compensation for awesome ChinAI contributors).
Feature Translation: Will AI Kill Visual China Group (视觉中国)
Context: Visual China Group (VCG), China’s leading provider of stock photography, has a bad reputation. VCG, which holds the copyrights for photos from Getty, AFP, Reuters, and BBC in China, is currently involved in over 10,000 legal disputes. Here’s how one venture capital firm partner described VCG’s business model in a Weibo post back in July 2018:
Visual China Group…have developed a system, by which they began to systematically search for various companies that use their pictures without authorization or with negligence. Then they ask for huge amounts of compensation. Usually, they do not accept deleting the image as a way to resolve a small act of negligence but instead directly ask for hundreds of thousands of RMB in compensation, and coerce the company to sign an annual contract. It really shouldn’t be the case, but infringement has become the core business form of this company now, which is also laughable. I don't believe that such a business model of blackmail can continue and be maintained. Just wait, one day...
Four years later, with advances in AI-generated images, has that day come? Let’s take a look at this article [link to original Chinese] by Cyzone(创业邦).
Key Takeaways: How did VCG get so big?
In the past, VCG’s customers were mainly professional media agencies and creative design companies. The explosive growth of WeChat public accounts (think: news portals and blog pages) has completely altered that landscape. The vast majority of articles I translate come from these public accounts. Today, there are almost 30 million WeChat public accounts!! Since these portals need images to spice up their text, image licensing has become a ten-billion-plus market.
Prices for VCG images can range from several hundred to several thousand RMB. An annual subscription can cost tens of thousands to hundreds of thousands. “Such a price is a necessary cost for large institutions; for small and medium-sized companies, it is a burden that cannot be afforded at all,” the article states.
VCG’s bad reputation is not just founded on high prices. Once, they tried to license the first-ever photo of a black hole, even though the image was freely available to all. The Communist Youth League of China has even criticized VCG for claiming a copyright of the national flag and emblem.
Can AI-generated images take down VCG?
Legal barriers should be surmountable. One cannot claim copyright for AI-generated works under Chinese copyright law, per an interview with a legal team from Dentons Beijing office,
There are, however, issues with generating precise images. Ran Nuochen, creative content director at ZCool, Chinese online platform for designers to share their work and build their portfolios, says:
At this current stage, there are still many flaws in AI painting works. Abstract and freehand works generated by AI may perform better, such as in the styles of Monet and Van Gogh, but when it comes to styles such as photos that require extremely high precision, AI-created works are prone to problems. For example, human facial features are wrong, or there will be an icon with an unknown shape in the lower right corner. These are all because AI does not have such strong discrimination when learning, and it will recognize the watermark on some training material pictures as the content of the picture itself.
One indicator that AI-generated images could challenge the dominance of sites like VCG is the launch of tools like JS Design’s “Instant AI [即时AI],” a tool built on Stability.ai’s Stable Diffusion model. Since its launch on September 15, it’s accumulated nearly 30,000 users, with favorable reviews.
FULL TRANSLATION: Will AI Kill Visual China Group (视觉中国)?
ChinAI Links (Four to Forward)
Must-read: The cost of China’s information vacuum
It’s been a long time since I’ve shared a must-read. This piece by Kathrin Hille for FT’s The Big Read starts by discussing how limited access to China shapes the understanding of Beijing’s decision-making by “those watching China from the outside.” This is an important topic, but I thought the most significant parts of the piece came in the piece’s second half, when it broaches how the current geopolitical environment contorts the demand for certain types of China-related scholarship:
Several China scholars say that recent administrations, Congress and big think-tanks are much less interested these days in more nuanced views that do not fit the narrative that Washington and Beijing are adversaries. One respondent to a NCUSCR survey published last year described research that is not hostile towards the other country as having “less reliable influence on policy and political debates” in both the US and China, according to Berris. “This is what psychology calls confirmation bias — you look for the answers you want to believe,” Lü says…But in the current political environment, it is really hard for scholars holding neutral views to exert influence because it seems regardless if it’s Republicans or Democrats, they are looking for hawks.” David McCourt, a sociologist at the University of California Davis who researches the community of “Chinawatchers” in the US, argues that the more hardline view about China also determines how younger scholars focus their research.
“If you’re wanting to position yourself as a future US policymaker, the horse has left the stable, engagement is now a dirty word, and so you’re either on board with strategic competition or you might miss out,” he says. He adds that when he initially intended to describe one group of experts still more in favour of continuing exchanges with China as “engagers”, some of the people were wary to even be identified with that term. “People don’t want to bear that label — they said you can call us the ‘responsible managers’ or the ‘responsible co-existers’, but just don’t call us engagers,” McCourt says.
Should-read: Governing artificial intelligence in China and the European Union
A group of Oxford Internet Institute researchers compare China and the EU’s AI strategies and goals. They study the mechanisms that the respective governments are using to promote the development and use of AI in the public and private sectors. Some very interesting takeaways on what the EU could lean from China’s approach to AI governance, especially regarding “out-of-control” AI.
Should-read: Understanding the Visual China Group Stock Photo Scandal
PlagiarismToday has a good background piece on past VCG stock photo scandals.
Should-read: This Uncensored AI Art Tool Can Generate Fantasies—and Nightmares
Last month, Will Knight, for Wired, dove into the world of “Unstable Diffusion, a Discord community for people using unrestricted versions of a recently released, open source AI image tool called Stable Diffusion.”
Thank you for reading and engaging.
These are Jeff Ding's (sometimes) weekly translations of Chinese-language musings on AI and related topics. Jeff is an Assistant Professor of Political Science at George Washington University.
Check out the archive of all past issues here & please subscribe here to support ChinAI under a Guardian/Wikipedia-style tipping model (everyone gets the same content but those who can pay for a subscription will support access for all).
Any suggestions or feedback? Let me know at chinainewsletter@gmail.com or on Twitter at @jjding99