ChatGPT competitor emerges - Anthropic's Claude - 분석맨의 '두 번째 뇌'

## My Thoughts - As a competitor to **[ChatGPT](https://openai.com/blog/chatgpt/)**, start-up **[Anthropic](https://www.anthropic.com/)** has launched an artificial intelligence system called **Claude**. - Still in closed beta, but receiving $700M in funding. - Opinions of those who tested the comparison are that Claude tries to answer in more detail than ChatGPT. - Claude was good in areas such as jokes, but overall ChatGPT still seemed to dominate. - ChatGPT's huge popularity makes it more likely that similar competitors or services will emerge. - ChatGPT is preparing for a paid version along with the paid version Professional, and it is expected that money will flow to the AI market as competitors emerge. ## Summary - Anthropic, a startup co-founded by former OpenAI employees, develops an AI system similar to OpenAI's ChatGPT. - The system, called Claude, is currently in closed beta. - [Related Twitter thread](https://twitter.com/AnthropicAI/status/1604929999743508480) - currently not accepting sign up - Created using a technology called Constitutional AI - Construct AI provides a “principle-based” approach to matching AI systems with human intent, allowing ChatGPT-like AI to answer questions using a simple set of principles as a guide. - The approximately 10 principles that shape this construct AI are based on the concepts of benevolence (maximizing positive impact), non-maliciousness (avoiding harmful advice), and autonomy (respecting freedom of choice). - Claude learned the probabilities of words occurring based on patterns, such as the semantic context of the surrounding text, from the huge number of text examples pulled from the web. - **OpenAI's ChatGPT vs. AnthropicAI's Claude Comparison** - Asked to Compare Himself to Machines from Science Fiction 'The Cyberiad' Asked to Write Poetry - [Related Twitter](https://twitter.com/goodside/status/1611556749726605312?ref_src=twsrc%5Etfw%7Ctwcamp%5Etweetembed%7Ctwterm%5E1611556749726605312%7Ctwgr%5E%7Ctwcon%5Es1_&ref_url=https%3A%2F%2Ftechcrunch.com%2F2023%2F01%2F09%2Fanthropics-claude-improves-on-chatgpt-but-still-suffers-from-limitations%2F) - Comparing Stanford AI Lab PhD Students - [Related Twitter](https://twitter.com/yanndubs/status/1611450984131354625) - Both are similar, but Claude tries to be more helpful, follows requests more closely, and is better at writing in English. Worst coding, worst French, longer response. - AI researcher Dan Elton's Claude vs. ChatGPT vs. Amazon Alexa Comparison - [Related Twitter](https://twitter.com/moreisdifferent/status/1612162231931920385?ref_src=twsrc%5Etfw%7Ctwcamp%5Etweetembed%7Ctwterm%5E1612162235471634439%7Ctwgr%5E%7Ctwcon%5Es2_&ref_url=https%3A%2F%2Ftechcrunch.com%2F2023%2F01%2F09%2Fanthropics-claude-improves-on-chatgpt-but-still-suffers-from-limitations%2F) - Claude jokes better than ChatGPT. - Claude is worse at math and coding than ChatGPT. ## References [Anthropic’s Claude improves on ChatGPT but still suffers from limitations | TechCrunch](https://techcrunch.com/2023/01/09/anthropics-claude-improves-on-chatgpt-but-still-suffers-from-limitations/) ## Links [[_MOC_2_AI-Robotics]]