Member-only story
How To Prep ChatGPT To Become Your Search Assistant Using Plugins
A step-by-step Python guide to adding 7 APIs that will enhance ChatGPT's knowledge of the outside world
What is the difference between search engines that use the Lucene open-source library and full-blown search engines such as Google? Inherently static queries are matched against individual words in a large index of content, while the other uses a complex combination of algorithms, machine learning models, and data analysis techniques to provide relevant search results.
Now, take Google and then add the power of conversational Chatbots powered by LLMs to the mix with access to the same search engines, and you have at your disposal a search engine that leverages a broadly trained NLP model capable of semantic searching where your search terms can derive similar text based on similar intent and meaning.
This is powerful even if you cannot define what you want clearly — an approximation of what you want can equally produce the desired results with greater ability.
LLM-based chatbots with the ability to search allow you to find data and interact with data directly from the confines of your chatbot interface.
In this article, we will look at how to add bridges to ChatGPT so your conversational chatbot experiences become more engaging with injections of dynamic real-time external search-based data sources.
Langchain will be our go-to wrapper to achieve the below design.
What Are We Going To Build?
While this is great to be able to add external search access into ChatGPT, there are some concerns you need to be aware of.
Given now, you will have access to various online data sources, noise becomes amplified, and information online could be more coherent and consistent depending on the source data is extracted from. You must be mindful of quality control issues or inappropriate responses, as the internet is free for all — a no-hold bar space.
