Scientists create ‘OpinionGPT’ to explore explicit human bias — and the public can test it By Cointelegraph

A team of researchers from Humboldt University of Berlin has developed a large language artificial intelligence (AI) model with the distinction of having been intentionally tuned to generate outputs with expressed bias.

Called OpinionGPT, the team’s model is a tuned variant of Meta’s Llama 2, an AI system similar in capability to OpenAI’s ChatGPT or Anthropic’s Claude 2.