Anthropic, a company founded by former members of OpenAI, is testing “Claude,” an AI chatbot similar to ChatGPT. Claude is built on a new, enormous model that was inspired by Anthropic’s prior 52 billion-parameter model, despite Anthropic being secretive about technical details.
Claude and ChatGPT both use reinforcement learning, but their training approaches are different. “Constitutional AI” is used by Anthropic to train preference models. Claude is available in two versions: Claude-v1, which excels in complex conversations and content creation, and Claude Instant, which works well for banter and text summaries. Notably, Claude AI demonstrated its abilities by succeeding on a test in law and economics while exceeding numerous humans.
User objects:
– Legal professionals
– Economists
– Researchers
– Content creators
– Students
– Casual chat users
– Professionals seeking document analysis and Q&A
>>> Please use: ChatGPT without login for Free