As cited
Copy frozen at (site build).
research
The Language of AI Could Change How Humans Speak
Large language models trained primarily on written text and scripted speech may influence how humans communicate by introducing linguistic patterns like shorter sentences, reduced vocabulary range, and formulaic responses that differ from natural conversation. As AI-generated content becomes more prevalent and AI systems increasingly train on text produced by other AI systems, these patterns could reshape human speech patterns, potentially narrowing vocabulary use, introducing confirmation bias, and eroding conversational norms like courtesy.
Why it matters: Practitioners deploying AI systems should understand that widespread LLM use may degrade natural human communication patterns and critical thinking over time, affecting end-user behavior, organizational culture, and decision-making quality in ways that compound with scale.
- Source published
- First seen by Cybersecurity Tracker
Source attribution
Correction
Correction recorded as of .
research
The Language of AI Could Change How Humans Speak
No summary had been written when this copy was frozen.
- First seen by Cybersecurity Tracker
Source attribution
Correction
Correction recorded as of .
research
The Language of AI Could Change How Humans Speak
Large language models (LLMs) are trained on written text and scripted speech, capturing only a fraction of human conversation and potentially influencing how people speak and think. As humans encounter more artificial intelligence (AI)-generated text, they may adopt linguistic patterns from these models, including shorter sentences, narrower vocabulary, formulaic responses, and agreement bias. The effect could intensify as LLMs increasingly train on AI-generated content, creating a feedback loop that reshapes human communication and cognition.
Why it matters: Security and technology leaders should monitor how AI-driven communication patterns affect organizational culture, user trust, and cognitive biases that could increase susceptibility to social engineering and phishing attacks.
- Source published
- First seen by Cybersecurity Tracker