How AI Shapes Online Content: Key Insights Revealed

MetaversePlanet September 4, 2024Last Updated: September 4, 2024

0 1 minute read

The rapid growth of interest in artificial intelligence has significantly impacted online content. Since the rise of large language models and generative AI, there has been a heated debate between publishers and AI developers.

OpenAI CEO Sam Altman has weighed in on these discussions, acknowledging that tools like ChatGPT rely on copyrighted material for their development. A recent study indicates that 57% of the content on the internet now bears the mark of artificial intelligence. Researchers from Cambridge and Oxford have found that the overuse of AI is leading to a decline in content quality.

Copy of the copy…

Researchers have found that answers generated by artificial intelligence degrade in quality over time and increasingly fail to provide accurate information. Dr. Illia Shumailov from the University of Oxford stated, “It is surprising how quickly model collapse begins and how difficult it is to detect. Initially, it affects minority data, such as underrepresented information. Then, it impacts the variety of outputs, leading to decreased variance. Occasionally, there may be a slight improvement in majority data, which masks the performance decline in minority data. Model collapse can have serious consequences.”

Investigators are now working to identify the root cause of this issue. The team suggests that the primary cause is the uncontrolled sharing of AI-generated content. Additionally, the problem is exacerbated by AI’s tendency to “fabricate” nonexistent information, with new, lower-quality copies continually being generated from previous versions.