Anthropic, the company behind the ChatGPT rival, Claude, has announced the release of an updated version of the chatbot. This new iteration promises more detailed answers, fewer harmful responses, improved reasoning, and overall better performance. Claude 2 is designed to be more conversational and human-like, making it easier for users to interact with.
Available in the United States and the UK, Claude 2 offers significant improvements over its predecessor. One notable enhancement is the increased processing power. The previous version could handle 9,000 tokens, whereas Claude 2 can now process 100,000 tokens. This upgrade allows the AI to understand and process over 70,000 words at once, resulting in more comprehensive and detailed responses. Anthropic claims that Claude 2 can process hundreds of pages of documents and generate longer outputs, including short stories.
Anthropic has also demonstrated the intelligence of its new AI model. Claude 2 achieved a score of 76.5% on the multiple-choice section of the bar exam, surpassing the previous version’s score of 73%. Additionally, the new version of Claude exhibits improved coding skills, with a jump from 56% to 71% on the Codex HumanEval, a Python coding test.
In terms of safety, Anthropic acknowledges that no model is completely immune from potential issues. However, the company asserts that it is now more challenging to prompt Claude 2 to generate offensive or dangerous responses. According to manual analysis conducted by Anthropic, the new AI model is twice as effective at providing safe responses.
Users can access Claude 2 through an API or via the dedicated site, claude.ai. Anthropic aims to provide a chatbot experience that is more advanced, reliable, and secure, catering to the needs of users in the United States and the UK.
– Photo Illustration by Rafael Henrique/SOPA Images/LightRocket via Getty Images
– AI Tool Stack