Anthropic, a startup founded by former research executives from OpenAI, has entered the AI arms race with the debut of its new AI chatbot, Claude 2. With a recent funding round of $750 million and support from companies like Google, Salesforce, and Zoom, Anthropic aims to develop advanced chatbot technology powered by generative artificial intelligence. Claude 2, which has been tested by businesses such as Slack, Notion, and Quora, is now being made available to consumers for the first time.
Anthropic’s focus has been on ensuring the safety and robustness of Claude. Daniela Amodei, co-founder of Anthropic, emphasized their commitment to creating the safest version of Claude yet. The company has been excited to offer access to a wider range of businesses and individual consumers. Initially, Claude 2 will only be accessible to users in the United States and the United Kingdom, with plans for expansion in the near future.
Generative AI chatbots, like OpenAI’s ChatGPT, have garnered significant attention and investment in recent years. They possess the ability to provide sophisticated and conversational responses to text prompts. However, concerns have been raised about the propagation of bias in such technology. Despite these concerns, generative AI chatbots have found applications in education, online travel, the medical industry, online advertising, and more.
OpenAI released GPT-4, a major update to ChatGPT, in March. This update allowed ChatGPT to browse the internet, expanding its knowledge beyond the 2021 training data cutoff. However, due to problematic uses, such as bypassing paywalls, OpenAI disabled the browsing feature. Google also introduced its AI chatbot, Bard, and has been continuously improving its capabilities.
Anthropic’s ambitions for Claude 2 are substantial. The chatbot can summarize up to approximately 75,000 words, equivalent to the length of a book. Users can input large data sets and request summaries in the form of memos, letters, or stories. In contrast, ChatGPT can handle about 3,000 words.
Daniela Amodei revealed that Anthropic invested at least two months in developing Claude 2, with a dedicated team of 30 to 35 people working directly on the AI model and a total of 150 people supporting it. The rapidly growing market for large language models presents ample opportunities for multiple players to succeed.
Anthropic’s achievements have earned recognition. In May, the company was invited to a meeting at the White House, along with Google parent Alphabet, Microsoft, and OpenAI, to discuss responsible AI development with Vice President Kamala Harris. Furthermore, Anthropic raised $450 million in funding, following a previous financing round of $300 million, valuing the company at $4.1 billion.
Claude 2 demonstrates notable improvements compared to its predecessor. It achieved a score of 76.5% on the Bar exam’s multiple-choice section, up from 73%, and improved its Python coding test score from 56% to 71%. In terms of safety, the upgraded chatbot showed twice the effectiveness in providing “harmless responses.” However, the challenge of AI chatbots generating incorrect answers, commonly referred to as “hallucinations,” remains.
Despite the advancements made, Daniela Amodei acknowledges that there are still significant hurdles to overcome in the industry. Language models, including Claude 2, are not entirely immune to hallucinations and face numerous unknown challenges. Nonetheless, Anthropic remains committed to pushing the boundaries of AI chatbot technology while prioritizing safety and reliability.
The rapid growth and demand for large language models present both opportunities and challenges for companies like Anthropic. While the market is expanding at an unprecedented rate, there are still unknown unknowns and known challenges that need to be addressed. One persistent issue in AI chatbots is the tendency to generate incorrect responses, also known as “hallucinations.” Anthropic recognizes this problem and acknowledges that Claude 2 is not immune to it.
To ensure the responsible development and deployment of AI technology, Anthropic has participated in discussions on AI ethics and responsible AI development. Their inclusion in a White House meeting with Vice President Kamala Harris highlights their commitment to addressing the ethical considerations surrounding AI.
The funding success of Anthropic, along with the investments in the AI industry as a whole, underscores the confidence and belief in the potential of AI chatbots. The significant financial support received by Anthropic demonstrates the industry’s recognition of their expertise and innovative approach.
