Google has announced a new feature for its search engine that will use generative artificial intelligence (AI) to provide more comprehensive and conversational answers to users' queries. The Search Generative Experience (SGE) feature is currently available in preview mode for users who sign up for the Search Labs program.
SGE is powered by large language models (LLMs) such as Pathways Language Model 2 (PaLM2) and Multitask Unified Model (MUM), which can generate natural language text from multiple sources of information. SGE can answer complex and multi-step questions by synthesizing a “snapshot” of relevant information from different webpages and presenting it at the top of the search results. Users can also ask follow-up questions to get more specific or detailed information.
For example, a user who wants to buy a laptop can ask SGE about the basic features to look for, and get a summary of different articles that contain the query. The user can then ask for laptop recommendations under a certain budget, and get a list of options from Google's Shopping Graph, which includes sponsored products. SGE can also help users with coding questions by using Code Tips, which can provide code snippets and insights for various programming languages. Here are some examples of how SGE can be used:
- You can use SGE to find more information about a specific topic. For example, if you're interested in learning about the history of the United States, you can use SGE to generate a list of relevant articles, videos, and images.
- You can use SGE to compare and contrast different products or services. For example, if you're looking for a new laptop, you can use SGE to generate a list of laptops that meet your criteria. You can then compare the different laptops to see which one is the best fit for your needs.
- You can use SGE to find answers to specific questions. For example, if you're not sure how to change a tire, you can use SGE to generate a list of step-by-step instructions.
AI is Still Improving and May Still be Inaccurate
SGE is an experimental feature that aims to make search more interactive and helpful for users. However, Google also acknowledges that generative AI is still a nascent technology that may have limitations and errors. SGE does not use a first-person response and avoids topics that may be controversial or sensitive, such as politics, health, or finance. Google also provides links to the original sources of information and encourages users to verify the accuracy and quality of the generated text.
Google's move to add generative AI to search comes amid the rising popularity of chatbots and conversational agents powered by similar technology, such as OpenAI's ChatGPT and Microsoft's Bing chatbot. Google hopes that SGE will enhance its search engine's capabilities and competitiveness in the AI era.
In recent months, we have seen the search landscape change, at least in terms of how tech companies envision the future of search. Microsoft took the lead with the launch of Bing Chat, its AI search chatbot that combines the companies own technology with ChatGPT-like capabilities from OpenAI‘s GPT-4 LLM.
Bing Chat has sparked curiosity about Microsoft's search engine, but it seems that people are not ready to switch from Google yet. I reported last month that Google was working its own AI search, powered by AI such as Bard and PaLM 2.
