A national survey by Elon University’s Imagining the Digital Future Center reveals that the adoption of artificial intelligence (AI) large language models (LLMs) such as ChatGPT is rapidly expanding in the U. S. , with half of Americans now using these technologies, marking one of the fastest adoptions in history. The full report is titled “Close Encounters of the AI Kind: The increasingly human-like way people are engaging with language models. ” The survey shows that LLM usage spans various demographics. While younger, wealthier, and more educated individuals are more likely to use LLMs, 53% of users hail from households earning less than $50, 000. Additionally, 66% of Hispanic adults and 57% of Black adults utilize these tools, compared to 47% of White adults. Users report experiencing human-like interactions with LLMs: 65% have engaged in conversations with back-and-forth dialogue (34% weekly), 49% perceive the models as smarter than themselves, 40% feel understood by their LLM, while 32% notice a sense of humor, and 25% think their LLM can make moral judgments. However, 50% of users also feel lazy with LLMs, and many express concerns about dependency and feelings of cheating or frustration. Lee Rainie, the center's director, emphasizes the surprising finding that personal, informal learning is the primary motive for using LLMs over work-related tasks; 51% use them for personal purposes. The survey indicates that 34% of users engage with LLMs daily, and ChatGPT is the most popular model, accessed by 72% of users. Regarding the impact of LLMs, 54% believe their productivity has improved, and 50% feel their ability to learn new skills has enhanced.

Users often recognize human-like traits in LLMs: 57% perceive confidence, 25% feel uplifted by the LLMs, and 22% sense empathy. Conversely, 23% have made significant mistakes relying on LLMs, and 21% felt manipulated. The survey also explored specific uses, showing that two-thirds utilize LLMs similarly to search engines, while many use them for brainstorming and planning. Looking forward, opinions are mixed about LLMs' future impact: 28% anticipate more positive than negative effects, while 59% expect job losses due to LLMs. Users predict substantial changes in leisure activities, work nature, and personal relationships with LLMs. In conclusion, as LLMs increasingly integrate into daily life, the findings provide a glimpse into the evolving relationship between humans and AI. The survey was conducted from January 21-23, 2025, among 500 language model users, with a margin of error of +/- 5. 1 percentage points.