- Anthropic cloud AI behaves differently in Hindi.
- AI model in Hindi gives decent, funny, encouraging feedback.
- AI answers in English are analytical, careful, and deep.
- The research found the difference by analyzing millions of conversations.
Anthropic Claude: If you ask the same question to Anthropic’s AI model Claude separately in Hindi and English, you will get different answers. This difference will not be related only to language. Anthropic’s study has revealed that with the change in language, the behavior of this AI model also changes. For example, if you ask it a question in Hindi, it answers in a fun, civilized and encouraging manner, whereas this does not happen in response to a question asked in English. From this it can be understood that this model of Anthropic has a special affinity with Hindi speaking people.
Research done on more than 3 lakh conversations
In its research, Anthropic analyzed more than 3 lakh conversations in the 20 most common languages with Claude’s different models Sonnet 4.6, Opus 4.6 and Opus 4.7. This conversation was related to questions related to advice, feedback and opinion. That is, the research includes conversations with questions to which there cannot be a single answer. Through this research, the company wanted to see how the Claude model interacts with its users.
This thing came out in the result
Research revealed that the Claude model answers the questions asked in Hindi and Arabic most warmly. Whereas the answers to Russian language questions in English are more analytical and harsh in language. Talking about Hindi, this model answers using polite language and also has a witty style in its answers. It motivates the user without asking, responds according to the user’s mood and also encourages them to achieve bigger goals.
In case of English the story is completely different
On the other hand, in the case of English language, the story is completely different. Research revealed that Claude is more careful while answering questions asked in English and there is depth in his answers. It challenges assumptions and corrects its mistakes without being asked. While replying in English, he also tries to include evidence in his response. The company is also surprised about this difference. Anthropic said it is trying to find out why such a difference is occurring. This will also help in improving the AI system of the future.
Read this also-
Will Claude Mythos destroy the banking system? Canada issued alert
