Google’s Search Generative Experience (SGE)

Why Trust Techopedia

What is Google’s Search Generative Experience?

Google Search Generative Experience (SGE) is an experimental pilot launched by Search Labs on 10 May 2023, which integrates generative AI alongside Google Search. This includes the ability to summarize search results, ask questions, and generate images.  


SGE is powered by multiple large language models (LLMs), including an advanced version of MUM and PaLM 2, which have been trained to carry out search-specific tasks. 

How Does Google’s Search Generative Experience Work?

As part of the Search Generative Experience, whenever a user enters a search into Google, they will be provided with an AI-generated summary of the results.

They will also be given the option to “ask a follow-up” or to select from a couple of other predefined questions to find out more information. 

When a user searches for a product, the SGE can give them a breakdown of products, product descriptions, reviews, ratings, prices, and images taken from Google Shopping Graph, which includes over 35 billion product listings. 

Users also have the option to use Google Search to create content, including texts and images. If a user enters a request to generate an image, the search engine will respond by creating and including up to four generated images in the results.  

AI-generated content can also be exported into Google Docs or Gmail. Clicking on an image reveals a description and the opportunity to edit the image. Users can also generate images directly in Google Images. 

How Can Users Access SGE?

Google’s Search Generative Experience is currently available to users in over 120 countries and territories.

You can activate SGE by going to the following Search Labs link and activating it via the toggle option. 

Users can also sign up via Search Labs in the Google app for Android and iOS or on Chrome desktop. 

A Look at Google’s Search Generative Experience Safety Guardrails

Google has developed the SGE with some basic guardrails to limit misuse and to keep users safe. This occurs by limiting the types of queries where generative search will be enabled:

  • For instance, SGE won’t produce output for searches around explicit or dangerous topics or those related to a “vulnerable situation,” such as a query about self-harm (in this case, Search will surface the contact information for a local support organization). 
  • In addition, SGE won’t produce automated responses if there’s a data void or information gap. This means it won’t respond if there is a lack of publicly available resources on a given topic, which could lead to low accuracy. 
  • It’s worth noting that Search also provides users with a warning message; “Generative AI is experimental. Info quality may vary,” as a basic warning about the potential for misinformation
  • When it comes to images, SGE will label AI-generated images with metadata and an embedded watermark to communicate that they were created by AI. This reduces the chance of deep fake-style scenarios, where users believe that false images are real. 

Why is SGE Important?

The inclusion of generative AI capabilities into Google Search highlights that generative AI is becoming increasingly accessible to consumers.

Going forward, selected users will have a search experience that is enriched by LLMs, surfacing and summarizing the information that’s most relevant to their original query. 

This builds on the existing capabilities of Google Search while also helping to differentiate Google from OpenAI’s popular chatbot, ChatGPT, combining search and generative AI as part of a single comprehensive solution. 

Above all, it will mean that users don’t need to search lists of content and drill down into individual blogs and articles to get access to insights. Instead, they can use AI-generated summaries to find the key information that matches their search intent. 

The Challenges Around SGE

While Google’s Search Generative Experience does have the potential to enhance search results for end users, it also introduces new risks. 

The most significant risk presented by SGE is that it could spread misinformation. Generative AI and LLMs are prone to hallucinations, which means they have a tendency to invent facts, figures, and other information. 

As a result, if a user enters a query into Google Search, there is the risk that the underlying LLM could misinform them by generating a prompt that promotes untrue information.

This means it’s important that users double-check the information provided by the summary to make sure that it’s legitimate. 


Related Questions

Related Terms

Tim Keary
Technology Specialist
Tim Keary
Technology Specialist

Tim Keary is a freelance technology writer and reporter covering AI, cybersecurity, and enterprise technology. Before joining Techopedia full-time in 2023, his work appeared on VentureBeat, Forbes Advisor, and other notable technology platforms, where he covered the latest trends and innovations in technology.