ChatGPT Finally Gains Web Search Capability
-
On September 28th (US time Wednesday), artificial intelligence research company OpenAI announced that their chatbot ChatGPT can now perform web searches using Microsoft's Bing, placing it on equal footing with Microsoft Bing and Google Bard.
Currently, OpenAI is striving to compete for dominance in the rapidly growing AI industry.
Microsoft and Google have already enabled their chatbots to conduct web searches and provide links to source materials to enhance the credibility and scope of information sharing. However, until now, ChatGPT has only been trained on time-limited datasets, restricted to information collected from the internet up to September 2021. In a series of tweets on Wednesday, OpenAI stated that ChatGPT can now utilize more up-to-date information.
OpenAI CEO Sam Altman tweeted, "We're back!"
Over the past few weeks, OpenAI, Google, Microsoft, Amazon, and Meta have been racing to release updates and new features for their generative AI products. Although many of these new tools still exhibit errors, this indicates the immense pressure these companies face to encourage users to generate new data and improve performance. On the same day OpenAI made its announcement, Meta also released a new chatbot that can perform web searches using Bing.
Previously, ChatGPT's paid users could enable web searches by linking browser plugins developed by other companies. The Wednesday update allows ChatGPT's core product to perform searches directly, potentially expanding its user base. Some users may also access this web search tool through a special testing program run by OpenAI.
Microsoft was the first to launch an internet-connected chatbot, introducing its Bing AI tool in February. Due to a multi-billion-dollar agreement with OpenAI, Bing incorporates OpenAI's technology. In March, Google launched its own chatbot, Bard, and added chatbot functionality to the top of search results for some users.
OpenAI has been rapidly developing consumer-facing tools, upgrading ChatGPT earlier this week to generate images and respond with voice, not just text.