The artificial intelligence (AI) chatbot named “Claude” from Anthropic, an organization created by former members of OpenAI and backed by Sundar Pichai-led Google will now compete with ChatGPT.
According to a blog post published on the company’s website, Anthropic’s study into developing helpful, trustworthy, and harmless AI systems led to Claude, a next-generation AI assistant.
Anthropic is co-founded by ex-Open AI employees Dario and Daniela Amodei and it recently received USD 400 million in funding from Google.
The new chatbot can perform various daily text-processing activities while retaining high dependability and predictability. It is accessible through the chat interface and API in the company’s developer console.
Claude can do the same tasks as ChatGPT such as writing blogs, summarizing text, replying to emails, coding, etc.
However, Anthropic claimed that Claude is ‘less likely to produce harmful outputs, easier to converse with, and more steerable’.
The chatbot’s tone, personality, and behaviour can also be tweaked to match the users’ requirements.
The statement said, “Claude can support use cases such as summarization, search, collaborative and creative writing, Q&A, coding, and more.”
The business released Claude and Claude Instant as two different software versions.
A high-performance model, Claude is available; however, Claude Instant is a lighter, less expensive, and significantly faster alternative.
The business also said it would introduce further upgrades in the upcoming weeks.
“As we continue to develop these systems, we’ll strive to make them more beneficial, trustworthy, and safe as we get more knowledge from our safety research and deployments,” Anthropic noted further.
Similar to ChatGPT, Claude cannot access the internet and has been trained on data till the 2021 spring.
However, Anthropic reportedly has taken a principled approach to its AI chatbot known as ‘Constitutional AI’. Claude has been trained on a large chunk of data that gives it the ability to avoid potentially dangerous topics based on its principles and even going to the extent of recognizing its own biases.
Before its launch, Claude had partnered with various companies to test out its technology. Quora is now using Claude to power its AI Chatbot app while DuckDuckGo is using Claude along with a combination of ChatGPT to summarize text from Wikipedia. Productivity app Notion is also using tech from Anthropic to power its Notion AI.
Claude also suffers from similar problems faced by ChatGPT and Microsoft Bing Chat such as “hallucinating” and users being able to bypass the chatbots’ safety features through clever commands.
News agency Bloomberg also quoted Anthropic CEO Dario Amodei, who admitted that their chatbot can sometimes make things similar to other language models.
He said, “I don’t want to say all the problems have been solved…I think all of these models, including ours, they sometimes hallucinate, they sometimes make things up.”