Anthropic, an artificial intelligence company established by ex-OpenAI scientists, has unveiled Claude 2, its latest chatbot designed to compete with ChatGPT and Google Bard. This upgraded model showcases notable enhancements, including an expanded token capacity, enhanced programming capabilities, and the introduction of a public beta website.
Described by its creators as a helpful and enthusiastic companion or personal assistant, Claude is positioned to excel in various tasks. The newly unveiled Claude 2 exhibits a significant leap in performance, as it can process up to 100,000 tokens per prompt. This capability enables Claude 2 to interact with content of extensive length, comparable to a book containing 75,000 words. These impressive token limits far exceed the previous constraint of 9,000 tokens in Claude 1.0, as well as the limits of 8,192 tokens in both Bard and ChatGPT4.
The enhanced capabilities of Claude 2 are demonstrated through its impressive performance in various tests. The chatbot achieved a notable score of 76.5 percent in the multiple-choice section of the Bar exam, showcasing its proficiency in legal knowledge. Additionally, Claude 2 excelled in the GRE reading and writing exams, achieving a ranking in the 90th percentile. Furthermore, the chatbot’s coding abilities proved to be advanced, as it attained a commendable score of 71.2 percent in a Python coding test.
Claude 2 goes beyond intelligent interactions and intricate tasks, placing a strong emphasis on safety. Anthropic has made significant efforts to enhance Claude’s safety measures, ensuring it is less likely to generate offensive or potentially harmful outputs. These improvements have been implemented within a short span of five months since the initial release of Claude 1.0, highlighting Anthropic’s commitment to continuously refining and prioritizing safety features.
The developers of Claude have implemented a distinctive “constitution” inspired by the UN Declaration of Human Rights and DeepMind’s Sparrow principles. This framework enables the chatbot to engage in self-improvement, recognize and address inappropriate behavior, and adapt its behavior autonomously, without the need for human intervention or feedback. By incorporating this set of rules, Claude aims to uphold ethical standards and ensure responsible conduct throughout its interactions.
Claude 2 extends its influence beyond its immediate practical applications, carrying significant implications for the realm of digital art and creation. The creative potential of AI, as exemplified by ChatGPT’s involvement in projects like the Turbo meme coin and lore development for Forgotten Runes, has showcased the capacity for AI to contribute to artistic endeavors. The introduction of Claude 2 represents a notable advancement in this trajectory of text, visual, and video-based AI artwork, ultimately benefiting the entire ecosystem involved in this creative domain.