Claude 2 Arrives! Anthropic Launches Powerful New Chatbot Competing With ChatGPT
Anthropic AI, which creates safer and more ethical generative AI systems, has launched Claude 2 - a new conversational AI chatbot competing with OpenAI's ChatGPT.
Available in the U.S and U.K in a public beta, Claude 2 improves on its predecessor in many ways according to early testers. Ethan Mollick, an Associate Professor at Wharton School, called out two advantages - it handles documents like PDFs well and shows a "sophisticated understanding" of content. Claude 2 also has a more "pleasant personality".
However, Mollick advised caution using Claude 2 for data. Though it can read CSV files, it still "hallucinates answers" unlike Code Interpreter.
Anthropic claims its "constitutional AI" approach makes Claude 2 less susceptible to harmful uses. "We have an internal red team evaluating Claude 2 on harmful prompts...while also manually checking results," the company said.
Compared to the first Claude, Claude 2 scored higher on the Bar exam questions, GRE reading/writing sections and Python coding tests. This comes after Google invested $300 million in Anthropic, purchasing a 10% stake and making Google Cloud Anthropic's preferred cloud provider.
Anthropic co-founders Daniela and Dario Amodei hope Claude 2 demonstrates how generative AI can become safer and more "steerable." The public beta launch marks a big step in that direction.