An AI requires vast amounts of data to be trained effectively. So far, training materials have been scraped from the internet, where most of human creativity is published. However, as we increasingly use AI to create new things, the internet will become saturated with AI-generated content. This content will then be used to train new AI models. Studies have shown that this process could lead to what is known as “model collapse” or “Habsburg AI.”

The Habsburg dynasty, one of Europe’s most powerful royal families, ruled over territories including Austria and Germany for centuries. However, their reluctance to share power led to intermarriage, inbreeding, and peculiar genetic defects. The famous “Habsburg jaw,” seen in portraits of Charles II of Spain by Carreno de Miranda, is just one example of the physical anomalies caused by inbreeding.

Similarly, AI models trained on extensive AI-generated material may produce strange and divergent results. The only way to avoid this is by incorporating fresh human-created content. But how can we achieve this when more people are using AI to create art instead of generating original work? AI-generated images are quick to produce, while human creativity is a slow and thoughtful process. Over the years, AI models have consumed thousands of years’ worth of human creativity and are still hungry for more.

It appears that human creativity will remain essential and in demand in the future. To prevent AI models from degenerating, we must infuse them with fresh human creativity. Human creativity may become a valuable resource, and those who can create something truly unique will be able to offer their services at a premium. There is hope that human creativity remains distinct and challenging to replicate, even in the age of AI.