At an Open Source AI meetup held at the Exploratorium in Downtown San Francisco on Friday, one of OpenAI’s key competitors unveiled their hotly-anticipated chatbot – Claude – to much hype and excitement in the crowd. Anthropic, a company backed by Alphabet and helmed by former OpenAI researchers, quietly opened up access to the new service to a limited number of people who had scanned a QR code that had been posted near the entrance.
With this exclusive rollout, those few select people got a chance to be among the first to get to try Claude, a chatbot interface that is brushing shoulders with ChatGPT. Although the official rollout of the service has been gradual, tens of thousands had quickly joined the waitlist after the business products released in early March. It has been a few weeks and access to the chatbot client Poe is open to the public, but access to Anthropic’s official Claude chat interface is still restricted. This event served as an opportunity for attendees to be granted access before the public roll out on March 14.
Anthropic stands out from other AI companies with its pioneering approach to AI safety through a technique called Constitutional AI. The concept involves providing a set of rules and principles for the system to follow and then training it using supervised and reinforcement learning methods. Claude has undergone safety evaluations and was deemed successful in its performance. Despite its successes, Claude, like all language models, is still susceptible to occasional ‘hallucinations’.
The concept of Constitutional AI shows a promise in advancing beneficial AI while avoiding any potential threats, paving the way to an entirely different AI landscape. Anthropic has already received funding from Open Philanthropy and other investors that are actively involved in the development of this revolutionary approach.
Anthropic is a company led by ex-OpenAI researchers that is backed by Google’s parent company, Alphabet. Founded in 2020, the company has made strides in developing groundbreaking techniques in artificial intelligence, particularly the concept of Constitutional AI. Through this approach, they aim to create an AI system that is in line with human intentions. With the backing of Open Philanthropy, they are working on making AI systems more helpful, safer, and more robust.
