Adapting to Google's Search Generative Engines (SGEs) - Part 2

How Adopting Emerging Search Generative Engines (SGEs) Gives Businesses a Strategic Advantage

From Bard and Bing to Perplexity, the search engine revolution is only beginning. My previous articles: " Adapting to Google's Search Generative Engines (SGEs)" which took a close look at Google's upcoming AGE, currently in Beta, and "Mastering SEO in the Era of Large Language Models", focused on the use of search engines by consumers and how to adapt SEO strategies.

Today, we look closely at some ways SGE capabilities can help organizations secure a strategic edge. Rather than jumping from one model to another, this article is based on Perplexity.ai's capabilities. Perplexity adds the power of ChatGPT4 Turbo and Claude2.1 to classic search functions and could be the model of future integrated search generative models.

Generating personalized search

For example, imagine that you are working in the AI field, your personalized Perplexity news feed will let you know that Microsoft is partnering with Mistral.

Clicking on that news bit, instead of redirecting you to a specific article, Perplexity provides you with:

1. selected sources, and links back to the sources.

2. A contextual summary of the news item

3. The option to ask follow-up questions

4. Related images

5. Suggested follow-up questions

   
   

Running all those searches manually would take a considerable amount of time, but the results are now available with a few clicks.

At this stage, checking the accuracy of the result by accessing the sources is still recommended to avoid potential hallucinations. Yet, even with that taken into account, the overall process is much faster as the sources are quoted and clickable.

Keeping up with specific competitors is also accelerated by creating a dedicated news feed.

Combining search with competitor intelligence

Even more impressive is the ability to generate a competitive intelligence report. Building on the example above, let’s imagine that you want to see how Mistral compares to other SGEs.

Just prompting Perplexity returns a comparative table copied below the screenshot to provide a better idea.

The full table generated by Perplexity is copied below to give an idea of how extensive it can be.

Of course, all these results need to be checked. Yet, as generating this table took less than 10 seconds, the time gained from finding and providing the information sources and organizing them in a comparative table is worth a try.

[1] https://mistral.ai

[2] https://blogs.microsoft.com/blog/2023/02/07/reinventing-search-with-a-new-ai-powered-microsoft-bing-and-edge-your-copilot-for-the-web/

[3] https://blog.google/technology/ai/bard-google-ai-search-updates/

[4] https://blog.google/technology/ai/google-gemini-update-sundar-pichai-2024/

[5] https://mistral.ai/technology/

[6] https://en.wikipedia.org/wiki/Microsoft_Bing

Etc.

Mining your data

SGEs can analyze and interpret unstructured data, like text documents, emails, social media posts, and images. Extracting insights from these sources today requires costly dedicated SaaS or software with limited flexibility.

With the growing number of tokens available to LLM (large language models) and SGE models, data analysts can converse with the model to fine-tune their search and uncover previously inaccessible insights. Currently, the best tool for that purpose is Perplexity.ai, which enables uploading a large amount of data and limiting the search exclusively to that data. That drastically reduces the risk of hallucination.

To provide an example, instead of feeding Perplexity with proprietary data, I asked it to "Provide a documented example of data mining based on publicly available data.

The question is: What are the population trends in terms of gender, life expectancy, and wealth inequality in U.S. states?"

Here is the answer it provided:

Data mining based on publicly available data can provide valuable insights into population trends. In the United States, recent data from the Census Bureau reveals interesting patterns in terms of population changes, gender distribution, life expectancy, and wealth inequality across different states.

Population Trends:

• Population Growth*: In 2023, more states in the U.S. experienced population growth compared to previous years[1][4]. States like Texas, Florida, North Carolina, and Georgia have shown significant population increases.*
• Population Decline*: Some states like Pennsylvania and West Virginia have faced population declines, although at a slower rate than before[1].*
• Migration Patterns*: Migration remains a key factor influencing population changes, with states like Texas attracting both domestic and international migrants[2][4].*
• Racial Diversity*: The demographic landscape is evolving with increasing diversity across states like California and Georgia[4].*

Gender Distribution:

• While specific gender data may not be explicitly mentioned in the search results provided, demographic trends often include gender breakdowns that can be analyzed to understand gender distribution patterns.

Life Expectancy:

• Life expectancy trends are not directly addressed in the search results. However, life expectancy data can be obtained from sources like the Centers for Disease Control and Prevention (CDC) or the World Health Organization (WHO) to analyze variations across different states.

Wealth Inequality:

• Wealth inequality is a complex issue influenced by various factors such as income distribution, access to education, and economic policies. Analyzing income data from sources like the U.S. Census Bureau or economic research institutions can provide insights into wealth distribution trends within states.

By leveraging publicly available data and employing data mining techniques, researchers can delve deeper into these population trends to gain a comprehensive understanding of gender distribution, life expectancy variations, and wealth inequality across U.S. states.

Citations:

[1]https://www.census.gov/newsroom/press-releases/2023/population-trends-return-to-pre-pandemic-norms.html

[2]https://www.census.gov/about/partners/sdc/projects/st-pop-trends.html

[3] https://www.commerce.gov/data-and-reports/population-statistics [4] https://www.visualcapitalist.com/population-of-us-states-compared-with-countries/ [5] https://www.pewtrusts.org/en/research-and-analysis/articles/2022/04/25/a-third-of-states-lost-population-in-2021

How can you use SGE/LLM?

These are just a few illustrated examples of how integrating SGE into organizational strategy can be a powerful accelerator.

Depending on the SGE/LLM specific capabilities, integrating with their available API can automate functions such as:

• Extracting insights from various data sources, including unstructured data like social media, emails, reports, images, audio, video, etc.
• Generating summaries and reports, condensing complex information into actionable insights, saving time and resources.
• Training bots based on unstructured data, knowledge base, analysts' reports, etc.
• Generating email funnels for different audience segments
• Personalize customer service based on prior interactions and query goal
• Personalize employee and customer onboarding and training process

And the list goes on and on.

Your imagination is the best way to find the model’s limits.

So, give it a go and let us know in the comments what other use you found.

Also, aside from creating a feed in our favorite SGE, the best way to keep updated with my content is to follow me 😀