Anthropic unveils Claude 2, a more advanced and less naive AI chatbot
Anthropic has released an upgraded version of its open-source ChatGPT competitor, Claude, a mere five months after its initial launch. This improved version boasts extended responses, enhanced justifications, reduced instances of hallucinations, and overall superior performance. Additionally, Claude now achieves a remarkable score in the 90th percentile among graduate school applicants on the GRE reading and writing tests.
An updated version, Claude 2, is available today for users in the US and UK. It can now handle up to 100,000 characters—that’s about 75,000 words or a few hundred pages of documents that users can have Claude condense and analyze—up from the previous version’s 9,000 character limit. In AI, credentials are the bits and pieces into which an input prompt is broken up so that the model can process them more easily—hence Claude’s ability to “digest” user data.
This increased capacity also results in longer and more nuanced responses. Claude 2 will even be able to generate short stories for “up to a few thousand tokens,” the company announced. Its coding capabilities have also improved, rising to 71.2% in the Codex HumanEval benchmark from 56%.
Claude’s “Constitutional AI” system is already guided by 10 secret “foundational” principles of justice and autonomy. Extensive testing by the red team since the release of the first version has made Claude 2 a more emotionally stable and harder to fool AI. Compared to its predecessor, Claude 2 is said to be “2x better at providing harmless responses compared to Claude 1.3,” the company’s release claimed. If you have already subscribed to the Claude 1.3 API, good news, you will be automatically upgraded to Claude 2 at no additional cost.
