AI Makes Internet Fake-Happy: New Study Reveals

A new study reveals that approximately 35 percent of all new websites, created between 2022 and 2025, are AI-generated or AI-assisted. This trend is contributing to an internet characterized by "sanitized and artificially cheerful" writing, according to researchers from Imperial College London, Stanford University, and the Internet Archive.

Analysis using sentiment analysis tools indicated that AI-generated or assisted content displayed a 107 percent higher positive sentiment score compared to human-written websites. This heightened cheerfulness is seen as a symptom of current large language models' optimistic nature.

Furthermore, the study found that AI-driven content is 33 percent more semantically similar, suggesting a reduction in the diversity of unique ideas and viewpoints online. Contrary to some expectations, the research did not find increased misinformation or a flattening of writing style.

Unexpectedly, the study's findings challenged common assumptions. Both researchers and the public polled had anticipated a rise in misinformation and a more generic writing style from AI. However, the data did not support these hypotheses, highlighting the unpredictable nature of AI's impact on the internet.