Concepts de base
Anthropic's Claude 2 is positioned as a safer AI bot, emphasizing accuracy and ethical responses in contrast to previous models.
Résumé
Anthropic's Claude 2, developed by ex-OpenAI engineers, aims to be a helpful and honest AI bot. Positioned as a safer alternative in the rapidly developing AI space, Claude 2 offers improved accuracy in coding, math, and reasoning. The model has been trained to avoid harmful responses and follows strict rules to ensure ethical behavior. Unlike other bots with controversial outputs, Claude 2 prioritizes safety and reliability while maintaining a human touch in its interactions.
Stats
Claude 2 scored 76% on the multiple-choice section of the 2021 multistate practice bar exam.
About 10% of Claude 2's training data was non-English.
Citations
"Perhaps we could have an interesting discussion about something else!" - Claude