The article also discusses the issue of AI chatbots "hallucinating", or making up their own facts. This is a common issue, with Google's LLM, PaLM-chat, inventing answers 27% of the time. The co-founders of Vectara, a software company specializing in semantic search, believe that chatbots could be trusted within a year or two when hallucinations will be close to zero. The name "Grok" comes from a 1961 novel by Robert A. Heinlein and means to understand something profoundly or intuitively.
Key takeaways:
- Elon Musk's company, xAI, has developed a chatbot named Grok, which is trained on real-time access to information from social media platform X.
- Grok's development is aimed at competing with other large language models like OpenAI’s ChatGPT, Google’s Bard, and Anthropic’s Claude.
- AI chatbots can 'hallucinate', or make up their own facts, with Google's LLM, PaLM-chat, having a high hallucination rate of 27% compared to ChatGPT4's 3%.
- The term 'grok', which means to understand something profoundly or intuitively, comes from Robert A. Heinlein's 1961 novel, Stranger in a Strange Land.