The “Dead Internet” theory is re-emerging in tandem with the rise of AI chatbots, according to a new article in The Atlantic that cites a paper published by UW researchers.
AI and chatbots are becoming more prevalent in communications, and users are starting to push back.
The paper, “Is This AI? Longitudinal Analysis of Strategies Used for AI Detection on Two Subreddits,” examines how people go about determining which material is AI-generated. Common methods include “fundamental plausibility” and “overly ornate but meaningless phrases.”
The Atlantic notes that, “The paper ends in an ominous tone. ‘Our findings suggest that people are developing mental models of what AI-generated content looks like.’ And that these mental models were unlikely to be accurate for long. The LLMs will evolve and stop doing the things that now make their output so easy to spot.
Tech Policy Lab alumna Tina Yeung closed the article by noting, “I definitely think this is a snapshot in time,” she said.
Read the full article here: The Dead Internet Theory is Back