Perplexity functions as an answer engine that uses AI language models like GPT-4 to summarize information collected by browsing the web. It retrieves relevant links, reads paragraphs, and generates concise answers for user queries. The system, known as retrieval augmented generation (RAG), aims to provide accurate information sourced from the web. Although perplexity minimizes hallucinations, it can still produce incorrect answers due to imperfections in the AI model. This raises concerns about the reliability and accuracy of AI search engines in providing correct responses to user inquiries.
A year ago, a chatbot tried to break up Kevin Roose’s marriage. Ever since, chatbots haven’t been the same. We’ll tell you how. Then, we’ll talk through the latest ways the world is adapting to artificial intelligence. And finally, Aravind Srinivas, the chief executive of Perplexity, will discuss his company’s “answer engine,” a challenger to Google’s search engine that could reshape the web as we know it.
Today’s guest:
- Aravind Srinivas, chief executive of Perplexity
Additional Reading:
We want to hear from you. Email us at hardfork@nytimes.com.
Find “Hard Fork” on YouTube and TikTok.
.