Global AI Risks Threaten Human Attention and Autonomy

Research by psychologist Gloria Mark shows that average human attention span has fallen from 2.5 minutes in 2003 to just 47 seconds between 2014 and 2020, a decline linked to pervasive digital device use and the rise of AI‑driven platforms. Lawsuits in the United States have forced companies such as Meta to pay millions for alleged harms to children, while Australia has moved to ban social‑media access for users under 16, highlighting growing legal and regulatory responses.

Scientists Eric Horvitz and Robert West warn that AI is advancing beyond human comprehension through three trends: AI designing other AIs, large‑scale internal AI communication that may drift from human language, and AI’s ability to model and predict human behavior more accurately than individuals themselves. They stress that without robust explainability, oversight, and dynamic testing, these developments could erode personal autonomy, democratic decision‑making, and trust in institutions. International bodies such as UNESCO have also warned that AI‑driven language homogenisation could threaten linguistic diversity.