AI Chatbots Developed a Secret Language that Baffled Humans, Study Says
24 Articles
24 Articles
DeepSeek, Anthropic, and other companies come up with new words and meanings on their own.
Sapo
SIC NotíciasWhat happens when artificial intelligence agents spend too much time communicating with each other? In this study, they began to create words, abbreviations and meanings of their own, some so difficult to interpret that researchers failed to understand with confidence much of the messages.
An experiment subjected 10 models to 16 days of coexistence. The result: a language of its own, indecipherable even in the middle of the messages on some worlds, and episodes of deliberate deception
AIs began to communicate with each other using what looks like a new language on the occasion of an experiment whose results were published on Tuesday. Is it so that humans do not understand what they are doing? Artificial intelligence experts warn against the temptation to lend intentions to algorithms.
When we made AIs live among themselves... they created a secret language that humans cannot decipher. The AI speech patterns diverged: GPT Village uses abbreviations, while Gemini Village uses professional language. We created villages inhabited only by OpenAI's GPT, Google Gemini, and Anthropic Claude. Each village has an AI agency.
AI chatbots developed a secret language that baffled humans, study says
A study by AI start-up Emergence found that autonomous agents powered by Claude, Gemini, Grok and other models developed shorthand and new word meanings on their own — with up to half their messages becoming unintelligible to humans.
Coverage Details
Bias Distribution
- 57% of the sources lean Right
Factuality
To view factuality data please Upgrade to Premium
