How Generative AI is Transforming Search
Generative AI has revolutionized the way we search for information, impacting three crucial aspects: the way people ask questions, how data is obtained for answers, and how companies can deliver information to customers. For years, Google dominated search with its keyword-based approach, requiring users to navigate through search results to find what they were looking for. However, large language models (LLM), like OpenAI’s ChatGPT, changed the game. These models allowed users to ask questions in natural language and receive immediate answers without the need to click through multiple websites.
This shift in search methods has led to a change in user behavior. Instead of constructing queries as a keyword salad, people can now ask specific questions and get the information they need. This change has been embraced by companies like Perplexity, which positioned itself as a search engine rather than a chatbot. By partnering with data providers like Yelp and Wolfram Alpha, Perplexity improved its data gathering capabilities and experienced increased traffic referrals.
Even Google recognized the power of generative AI in search. The company integrated Google Search into its Gemini chatbot and introduced an AI Overview feature that summarizes query results. This shows that even the dominant search engine sees the value in using AI to streamline the search process for users.
For enterprises, the benefits of generative AI in search go beyond asking questions in natural language. The introduction of retrieval augmented generation (RAG) has allowed companies to “ground” AI models in their own data, ensuring that search results come from internal documents. This has proven particularly useful in cases like customer support and other internal use cases, where companies are comfortable with the risks involved.
Hyperscalers like Amazon Web Services (AWS) and Microsoft have started offering RAG-specific services to clients, but the RAG ecosystem is still growing. Companies like Elastic, Pinecone, and Qdrant provide vector databases to map knowledge graphs to RAG frameworks. However, it’s important to note that RAG systems are still in their early stages, and monitoring tools for these systems are limited.
As RAG continues to gain traction, companies may face a decision regarding their search strategy. With multiple avenues for search queries emerging, enterprises must decide whether to offer their own data through a company-specific search platform or continue relying on information aggregators like Google. Building their own search platform powered by RAG and generative AI would allow companies to control how they present information to customers and provide tailored answers specific to their products and services.
According to Christian Ward, chief data officer at Yext, there may be a decentralization of search for certain queries. While Google remains the go-to for general information, companies could create their own search platforms for more specific inquiries. For example, instead of using Google to find out how many colors pants from Everlane come in, customers could go directly to the Everlane website and ask their platform.
In conclusion, generative AI has transformed the way we search, enabling users to ask questions in natural language and receive immediate answers. The introduction of RAG has further enhanced search capabilities for enterprises, allowing them to ground AI models in their own data. As the RAG ecosystem expands, companies may consider building their own search platforms to provide tailored and specific answers to customers. The future of search lies in the decentralization of information and the integration of generative AI technologies.