Guide to Google Generative AI in Search (SGE)

Ready to talk? We'd love to help.

person with search engine open on laptop at a business meeting

Understanding Generative AI in Search (SGE)

The surge of Artificial Intelligence (AI) in our everyday lives has been hard to ignore. Delving into the depths of AI’s interaction with the World Wide Web, Google has unveiled a new major change: Generative AI in Search (SGE).

This shift in search is set to redefine the user experience, making it more informative, intuitive, and interactive. But what is SGE, how does it work, and what are its potential limitations? Read on to find out!


A Historical Context of AI in Search

The concept of AI isn’t new. Pioneered by visionary figures like Alan Turing, it’s evolved significantly since its inception. In the past, AI was more of a table talk for scientists and enthusiasts.

Fast forward to 2023, there was an unprecedented surge in AI’s usage and recognition. Most notably, Open AI’s ChatGPT launched in November 2022. The platform gained 100 million users in just two months, making it the fastest-growing consumer app ever.

Consultancy PwC estimated that AI-related economic impact could reach $15.7 trillion globally by 2030 – almost the gross domestic output of China. With new AI platforms and technologies seemingly being released every day, it’s no surprise that it will have a major impact on how we search.

Google Search is transforming under the influence of AI, with SGE being a prime example. Let’s dive in and find out what this could mean for you and your business.


What is SGE?

SGE, the latest innovation in the world of search, is the marriage between generative AI and Google’s search capabilities. But what is SGE?

Simply put, it’s an enhancement of Google’s existing search functionality. It allows users to explore complex queries, obtain comprehensive overviews of their desired topics, draft documents, and even create visual content directly from a single search.

With generative AI in search, you can:

  • Ask more descriptive, complex questions
  • Understand a searched topic more quickly with relevant links
  • Generate imagery or write drafts while searching
  • Ask for follow-ups in a conversational tone
  • Suggest next steps

Essentially, Google’s search results will be organized in a new way to help users find exactly what they need with a single search query. Here’s how it will look:


screenshot of an example of generative ai in search from google

Source: Google

So, how does it all work?


How SGE Works

Generative AI operates on the principle of creating something new based on a given set of parameters. In the context of SGE, the AI system is designed to understand the user’s query – but then go a few steps further.

SGE leverages its understanding of language, context, and the user’s intent to deliver multifaceted responses. Several components work together to achieve this:

  • A Language Model to comprehend and generate natural language text
  • A Multimodal Integration for incorporating and creating multimedia experiences
  • A Task Representation that manages the workflow and the feedback loop for the user

SGE is set to revolutionize the search experience by introducing AI-powered overviews for quick, comprehensive snapshots on a variety of topics. Supported by links to authoritative resources, users can verify the information provided and discover more from publishers, creators, retailers, and businesses.

For a more dynamic search experience, SGE also simplifies follow-up queries. Users can easily ask follow-up questions without repeating their initial query. The result is a more fluid, conversational engagement with search.

Generative AI also plays a crucial role in vertical-specific searches such as shopping or local inquiries. In shopping searches, generative AI provides snapshots that highlight important product information like reviews, prices, and images.

Likewise, for local searches, AI-powered insights offer comparative views on places, helping users make informed decisions quickly.



Using Generative AI Ethically

The power that AI offers also carries significant ethical implications. These implications have already been acknowledged with proposed policies such as the EU AI Act and the AI Bill of Rights.

The rollout of SGE is experimental and comes with a strategy:

  • Ethical deployment and ongoing improvements: Google’s advancement process involves collaborations with industry specialists, policymakers, civil rights leaders, content creators, and more.
  • Mitigating LLM limitations: By restricting SGE to specific tasks and incorporating corroborative functionalities, Google substantially reduces common Large Language Model (LLM) issues like hallucination.
  • Human oversight: Human feedback is vital for optimizing SGE’s outputs. Key focuses are on factors such as clarity, length, and format.
  • Quality and safety in search: SGE integrates with Google’s core search ranking and quality systems. This ensures that SGE maintains reliable and helpful standards, especially on sensitive or critical topics, designated as “Your Money or Your Life” (YMYL).
  • Policies for safe, accurate information: SGE adheres to stringent policies that prevent the generation or display of harmful content. This includes adding safeguards for teens and incorporating feedback mechanisms to continuously improve user protection.

Data privacy, cybersecurity, and the impact on the workforce are all valid concerns that require ongoing attention and regulation. The focus is on technology companies developing AI to maintain open and honest conversations with their users.


Known Limitations of Generative AI in Search

No technology is without its imperfections, and generative AI is no exception. SGE is still evolving. It faces challenges in understanding nuanced language and managing context shifts. At times, it can also generate irrelevant or inaccurate outputs.

Despite the strides made with SGE, there are some limitations:

  • Misinterpretations during corroboration: There are instances where SGE correctly identifies information for corroboration but fails in its interpretation, leading to outputs that slightly misconstrue the intended message.
  • Hallucinations: SGE may distort facts or inaccurately reflect insights, a phenomenon often referred to as a “hallucination.”
  • Bias: Training on high-quality, web-extracted data does not shield SGE from inherent biases. These may manifest in narrow representations or undesirable associations, despite Google’s efforts to implement bias-mitigation measures.
  • Opinionated content: While designed to maintain a neutral stance, SGE can inadvertently reflect existing web opinions.
  • Duplication or contradiction: Given its integration with existing search features, SGE’s outputs may sometimes duplicate or contradict information presented through other search functionalities.

Recognizing these limitations, Google has expressed their commitment to ongoing improvements through model updates, fine-tuning, and a dedicated effort to address bias and enhance the reliability of SGE’s outputs.


Stay Ahead of the Future of Search

Adapting to this technological wave can be overwhelming, especially for businesses trying to optimize their online presence. That’s why having an expert partner like BNP Engage can be your key differentiator when understanding the complexities of AI search marketing strategies.

With a wealth of experience in SEO, AI, and all things digital, BNP Engage is ready to be your guide. Our team can help ensure your website is optimized to take advantage of the latest changes. Contact us today to get started!

Lauren DevensApril 11, 2024by Lauren Devens