Baidu, a leading technology company, has just presented ERNE-ViLG, a new text-to-image AI capable of creating realistic representations of the Asian giant. It is also capable of improving other leading AIs such as DALL-E 2.
What does not change is censorship in china. It is that, for example, the second largest square in the country and with a great historical, cultural, and political load, Tiananmen Square, does not appear in this AI tool. This is a fact that users did not notice when a demo of this software was released at the end of August.
Quickly, they noticed that certain political leaders or words considered controversial were labeled as "sensitive." So they were blocked in the results. China's 'competitor' DALL-E 2 prohibits sexual content or images of medical personnel.
The ERNIE-ViLG model is part of Wenxin, a large-scale project in natural language processing from Baidu. It is made up of a data series of 145 million image-text pairs and contains 10,000 million parameters that the AI uses to distinguish subtle differences between concepts and art styles.
What does this mean in practice? That ERNIE-ViLG has a smaller data set than DALL-E 2, which remains at 650 million pairs, and Stable Diffusion, with 2.3 billion pairs. But, nevertheless, more parameters than either of the two. At the moment, Baidu has released a demo version on its own platform as well as on Hugging Face, one of the most important international AI communities.
The main difference between ERNIE-ViLG and other western software is that Baidu's software understands written instructions in Chinese. Also, you are less likely to make mistakes dealing with such culturally specific words.
A Chinese video creator compared the results of several models that included Chinese historical figures, pop culture celebrities, and food. And he found that ERNIE-ViLG produced more accurate images than DALL-E 2 or Stable Diffusion. After its release, ERNIE-ViLG has also been adopted by the Japanese anime community.
In a study conducted by the MIT Technology Review, it was discovered that several Chinese words had been blocked as names of high-profile Chinese political leaders like Xi Jinping and Mao Zedong. Also terms that may be considered politically sensitive, such as'revolution', or the name of Baidu's founder and CEO, Yanhong (Robin) Li.
Words like 'democracy' and 'government' are permitted, but this results in the use of other words, such as 'democracy in the Middle East' or 'British government.'
It's not a new concept, far from it. In China, social media companies often have lists of sensitive words that have even been created by government orders.
The Chinese government proposed a new rule that would prohibit any AI-generated content that, according to him, jeopardizes national security and social stability in January of this year. ERNIE-ViLG is an excellent example of how this organization can continue to play an important role in the future of large-scale text-to-image AI.