Google and OpenAI are engaged in fierce competition in search functionality. Google has launched Gemini, integrating Google Search to provide more accurate answers; while OpenAI quickly rolled out ChatGPT Search, allowing users to access real-time web data directly, including sports scores and news updates. ChatGPT is capable of understanding context, providing more coherent responses, and collaborating with news and data providers to enhance the visualization and timeliness of information. This feature will gradually be made available to Plus and Team users, and will also be rolled out to all free users in the coming months
Just now, Google announced with great fanfare that Gemini integrates Google Search, giving it the ability to "Ground." In less than a minute (one at 10:00 AM and the other at 10:01 AM), OpenAI immediately announced the official launch of the ChatGPT search function! The timing is so precise that it’s hard to believe it’s a coincidence; is there a mole inside Google? The air is thick with tension!
First, let’s take a look at ChatGPT Search: this time it’s not just a simple addition of a search box, but a deep integration!
Now ChatGPT can directly access real-time web data, including sports scores, news updates, stock price fluctuations, weather forecasts, and restaurant recommendations! Previously, users had to jump to a search engine, but now they can search directly in the chat box!
OpenAI also emphasized its collaboration with news and data providers, offering a brand new visual design and richer real-time information covering multiple areas such as weather, stocks, sports, news, and maps. Users can directly click the source button to open a sidebar to view reference links.
What’s smarter is that ChatGPT can "understand" your questions and intelligently determine whether to initiate a search. Of course, you can also manually click the web search icon to control the search pace yourself. More importantly, ChatGPT will consider the context of the entire conversation to understand your questions better, providing more accurate and coherent answers.
This web search feature is initially available to ChatGPT Plus and Team users, as well as users on the SearchGPT waitlist. Enterprise and education users will gain access in the coming weeks, and all free users will gradually have access in the coming months. It will be available on both the web version and mobile applications.
Now let’s take a look at Google Gemini. Google emphasizes that by integrating Google Search, Gemini can provide more accurate and up-to-date answers, along with source links and search suggestions to enhance transparency and credibility.
Google has given this feature a grand name, called "Grounding with Google Search," which essentially means that the Gemini model can connect to Google Search in real-time to obtain the freshest and most comprehensive information In the future, when asking it questions, you no longer have to worry about it spouting nonsense seriously or giving outdated answers.
A few examples were given. For instance, asking it "What are the new CSS features this year?" the online Gemini can accurately list the new features for 2024 and even thoughtfully provide source links, full of details!
Another example is asking it "What is the smallest horse breed in the world?" The online Gemini not only knows the correct answer is the Falabella mini horse but can also explain the difference between breed size and individual size in detail, even mentioning those mini horses that are smaller than Falabella in the Guinness World Records.
To avoid the need to search online for every question, wasting time and money, Google has also developed a "dynamic retrieval" feature. It intelligently determines whether an online search is necessary based on your question, and you can adjust the dynamic coefficient yourself.
In terms of technical details, OpenAI revealed that the search function of ChatGPT is based on a fine-tuned GPT-4 model and uses a brand new synthetic data generation technology, including outputs extracted from OpenAI's o1 preview version.
The search function of Gemini is based on the Gemini 1.5 series model, and developers can use the google_search_retrieval tool via API or enable it in Google AI Studio. In AI Studio, this feature is available for free testing, while in the API, it requires payment, priced at $35 per 1,000 queries. Is this price serious?
This marks the official challenge from OpenAI to Google Search, but currently, I feel that the AI search experience from both tech giants is not as good as Perplexity. However, the big companies have money and talent, and the competition in AI search has just begun Author of this article: opencat, Source: AI Cambrian, Original title: "A Hard Hit! One Minute After Google Gemini Launched Search, OpenAI Quickly Responds with the Official Launch of ChatGPT Search"