The images in this blog are now almost all generated by AI. Perhaps it is over 95% of the images. The older articles use mainly photos, but articles from the last few months are mostly AI-generated. Let’s talk about the AI-generated images.
The discomfort of AI-generated human expressions
I had a question. It is why several kinds of AI-generated human expressions look so strange.
Now, almost all AI-generated images look more attractive than the actual photo. I have compared them over the past months to choose the pictures for this blog, and that made me feel that way. Cakes look more delicious, and flowers are more vibrant. AI does great work.
Perhaps AI modifies them to be more impressive. It emphasizes these characteristics, such as gloss and colors. That enhances their attractions.
However, human expressions remained strange. The typical example is the shape of the mouth called duck face or duck lip. Look at the following images:
In my perception, I frequently saw images with such a relatively large mouth with emphasized smiles. They made me uncomfortable.
That made me wonder why.
Perhaps there is a difference in the universal sense of the human face. In other words, there is a difference in sensitivity to charm. Today, I will talk about the logic.
Why AI emphasizes the mouth
In my opinion, we pay attention to different facial features based on personality and culture. In other words, it is not the fault of AI. The difference in human preference causes it.
In the case of the mouth shape mentioned above, the following two elements would be the cause:
- Difference based on personality: Less empathic people focus more on the mouth, while highly empathic ones focus on the eyes.
- Difference based on language: Languages that focus on consonants, such as English, pay attention to the mouth rather than the ones that focus on vowels.
People emphasize the parts they pay attention to when they illustrate them. This creates the differing sensitivity to charm.
Let’s look at each below.
Less empathy vs. high empathy
First, people pay attention to different places based on their personality. Less empathic people focus more on the mouth, while highly empathic ones focus on the eyes.
You can easily understand this tendency by watching cartoons.
For example, characters with high empathy have large eyes. A typical example is illustrations for young girls. These main characters have bigger eyes and smaller mouths. Adolescent girls prioritize empathy the most during this period. These characters express emotions with their eyes and subtle facial nuances.
On the other hand, the characters in less empathic stories have smaller eyes and bigger mouths. The more insane the story is, the narrower the eyes become. In other words, less empathic people judge the opponent’s emotions by mouth and words.
The less empathic people are the social majority. That would be why highly empathic people tend to feel uncomfortable with the illustration of a person with a larger mouth.
The difference in language
Second, languages can affect people’s attention to the mouth. Languages that focus on consonants, such as English, pay attention to the mouth rather than the ones that focus on vowels.
People whose mother language is not English often notice something when they start learning English. Speakers in movies and television programs emphasize the shape of the mouth. They move their mouths extremely dynamically.
This would be because consonants are more difficult to hear than vowels, and they can accurately confirm consonants through the shape of the mouth. In other words, watching the mouth is a part of listening.
On the other hand, in languages that emphasize vowels, it is unnecessary to see the mouth. Words can be inferred from the combination of vowels. Examples are Japanese, Korean, and Chinese.
You can also see this difference by comparing cartoons from these regions.
Cartoons made in the U.S. and the U.K. have larger mouths and are particular about lip sync. On the other hand, East Asian cartoons have smaller ones.
The sense of charm
These elements cause the difference in the sense of charm.
It means that highly empathic people whose mother tongue is other than English prefer big eyes and a small mouth the most. On the other hand, less empathic native English speakers tend to like a bigger mouth.
When we describe emotions with our mouths, we smile slightly at the mouth corners. That creates that strange, long mouth with an uncomfortable, stereotypical smile.
In other words, that shape of the mouth is the symbol that low-empathy people find charming.
It means that that uncomfortable tendency will continue, no matter how much AI improves. They are not the fault of AI. The difference in human preference causes it.
Conclusion
That is the logic of why AI creates such strange human expressions.
We pay attention to different facial features based on personality and culture. The shape of the mouth would be the symbol that low-empathy people prefer.
This might resolve our discomfort with AI-generated images.
Thank you for reading this article. I hope to see you in the next one.
