Google Launches New AI Search Engine: How to Sign Up
Google has launched Search Generative Experience, or SGE, an experimental version of Search that integrates artificial intelligence answers directly into results, the company said in a blog post Thursday.
Unlike a normal Google Search, which brings up a list of blue links, SGE uses AI to answer your questions directly on the Google Search website. After entering a query into Google Search, a green or blue box will expand with a new answer generated by Google’s large language model, like the one that powers OpenAI’s ChatGPT.
Google pulls this information from the websites and links to the resources used when generating an answer. It is also possible to ask follow-up questions in the SGE to get more accurate results.
At this time, SGE is not open to the public and requires you to register with Google Search Labs. To join, click the link here. The research labs are currently only available to a limited number of people in the US and only in English, although you can join the waiting list. Google did not immediately respond to a request for comment.
With the launch of ChatGPT late last year, an AI chatbot that can answer almost any question with a unique answer, companies have been adding AI-generating features to their products amid increased public interest. Google unveiled Bard earlier this year, an AI chatbot similar to ChatGPT. Microsoft went on to add ChatGPT to Bing directly, including an AI image generator powered by Dall-E, also from OpenAI. AI chatbots are powered by a large language model, or LLM, a technology that uses a massive data set of text to write sentences that mimic human language. The model essentially aims to figure out what the next best word should be when generating sentences, a process that has been described as “autocompletion on steroids.”
AI was also a key focus earlier this month at Google I/O, the search giant’s annual developer conference, with the term being mentioned more than 140 times during the two-hour presentation. During I/O, Cathy Edwards, vice president of engineering at Google, said that with a standard Google Search, people have to break down complex queries into multiple queries, scan web pages for information, and formulate the answer in their head. With SGE, AI can do all of this for you.
How to join the Google Search Labs waiting list
Here’s how to join the Research Labs waiting list so you can be among the first to test Google’s SGE:
Open the Chrome browser on a computer. Sign in to your Google account. Open a new tab in your browser. In the top right, there will be a Labs icon (of a glass) if Labs is available to you. If the Labs icon is there, click it and then click Join Waitlist.
You will receive an email when Labs are available.
SGE is part of Research Labs and includes experimental features such as Code Hints, which provides coding suggestions directly in Search, and Add to Sheet, a feature that can automatically bring information found in Search to ” Google Sheets”.
If you are able to access SGE now, Google requires you to accept its privacy notice and asks you not to include sensitive or confidential personal information that “can be used to identify you or others in your interactions with the Features of SGE”. This is because during this testing, some data will be analyzed by human reviewers, although the data will be “stored in a way that is not associated with your Google account”. It is possible to delete interactions via the My Activity page.
Google also warns about AI generating nonsense and that accuracy may vary. This likely refers to “hallucinations”, a problem found with generative AI where it can confidently say something is correct when it isn’t. Google suggests that you not rely on generative AI for medical, legal, financial or other professional services.
Editors’ note: CNET is using an AI engine to create some personal finance explainers that have been edited and vetted by our editors. For more, see this post. SGE can be accessed via the Chome desktop web browser or the Google Android and iOS apps. a large language pattern