Why Doesn't AI Want to Show Me Jesus Washing the Feet of His Disciples?

🌈 Abstract

The article discusses the author's inability to generate an image of Jesus washing the feet of his disciples using various AI image generators. The author explores possible reasons for this, such as the generators potentially having restrictions on depicting Jesus in a subservient position, or simply not having enough training data on this specific biblical scene.

🙋 Q&A

[01] Why Doesn't AI Want to Show Me Jesus Washing the Feet of His Disciples?

1. What is the author's main observation regarding their attempts to generate an image of Jesus washing the feet of his disciples? The author notes that they were unable to generate such an image using various AI image generators, despite this being a common and well-known biblical scene that has been frequently depicted in Christian art.

2. What are some of the possible reasons the author suggests for the AI's apparent reluctance to generate this image? The author proposes a few potential explanations:

• The AI generators may have restrictions or parameters in place that prevent them from depicting Jesus in a subservient position, as the act of washing feet could be seen as.
• The AI may be biased towards generating images where Jesus is being served or having his feet washed, rather than the reverse scenario.
• The AI may simply not have enough training data on this specific biblical scene, despite its prominence in Christian art and mythology.

3. How does the author's experience with generating other images, such as a triangle, factor into their analysis? The author notes that they had similar difficulties generating other specific images, such as a guy staring off the bow of a sinking, burning ship. This suggests that the AI generators may struggle with producing certain types of detailed or specific images, even if the subject matter is not particularly controversial or sensitive.

[02] Comparison of AI Image Generation Results

1. What are the key differences the author observes in the results from the various AI image generators they tried? The author notes that the results from different generators, such as Bing's DALL-E 3, Imagine.ART, Stable Diffusion 2.1, Craiyon, and DALL-E mini, all had varying degrees of success in depicting the requested scene. Some generated images where Jesus was having his feet washed, rather than washing the feet of his disciples.

2. How does the author interpret the differences in the generated images? The author suggests that the differences in the generated images may be due to the AI's potential biases or restrictions, as well as the varying levels of training data available to each generator. The author wonders if the AI may be hesitant to depict Jesus in a subservient position, even though this is a significant biblical scene.

3. What does the author's experience with trying to generate a simple image, such as a triangle, suggest about the overall capabilities of these AI image generators? The author's difficulty in generating a simple image of a triangle, despite the apparent simplicity of the request, suggests that these AI image generators may still have limitations in producing certain types of specific or detailed images, even if the subject matter is not particularly complex or sensitive.