Agreeing With Algolia- ChatGPT is Great but Search is Here to Stay

Since its release in November 2022, critics have crowned ChatGPT the newborn King of search. Everyone cited OpenAI’s brilliance, and rightly so. However, calling this chatbot the prime candidate to overthrow the existing search engine monarch is where we draw the line. Barely a month after the release of ChatGPT, Bloomberg’s review of ChatGPT spelled doom for Google… Kindly pause for laughter.

Algolia to Google’s Defense

Algolia took their time and released a rebuttal, albeit to no one in particular. Essentially, they come to the rescue of search engines. This is not unexpected; after all, Algolia is predominantly a search and discovery company. However, is there some truth to their article about the need for search engines after the emergence of ChatGPT? ~~Spoiler alert: There is.~~

In Algolia’s article, the author explains the technology behind search in general. The concept of query formulation (how information is requested in a specific syntax) is key to understanding the writer’s stance. ChatGPT consistently fails to identify and correct common syntax mistakes made by users. Google and Kagi are quick to realize that these are errors. Also, for some purposes, traditional search engines display results in a nice interface that gives users variety. This variety can be beneficial in some situations.

ChatGPT excels at a lot of things because of the underlying technology of chatbots. However, it works differently from Google and other search engines in general. This difference makes each platform unique in its own way. Not bad, not good, just different. Very quickly, here’s how ChatGPT works, and we juxtapose it with how Google works.

How ChatGPT Works and How Google Search Works

GPT-3 (short for “Generative Pre-training Transformer 3”) s a neural network-based language model that uses machine learning techniques to generate human-like text. It uses a technique called “transformer-based language modeling,” which involves processing the input text through a series of interconnected “layers” or “transformers.” Each transformer takes in a sequential order of words and outputs a new sequence of words to capture the relationships between the words and their meanings.

How Google Search Works

There are four essential parts of Google search engine.

Crawling

Google uses automated software called “spiders” or “bots” to discover new web pages and follow links from one page to another. This process is known as “crawling.” As the spiders crawl the web, they create a copy of each page they visit and add it to Google’s index, a massive database of all the web pages that Google has discovered.

###Indexing Once a page is in Google’s index, it can be included in search results. However, not all pages are indexed. Google uses algorithms to determine which pages are the most relevant and useful, and those are the ones that are more likely to be included in the index.

###Ranking After entering a search query, Google’s algorithms analyze the index to find pages that are relevant to the query and rank them in order of relevance. Google uses hundreds of factors to determine the relevance and ranking of web pages, including the page’s content, the quality and relevance of external links pointing to the page, and the user’s search history.

Results

Google displays the search results to the user, with the most relevant results appearing at the top. The results page also includes sponsored results, which are paid advertisements ranked based on a combination of relevance and the bid amount.

This is a significantly simplified explanation, but you get the point.

Why We Agree with Algolia - Search is Here to Stay

As an experiment, we put ChatGPT to the test in three ways.

Speaking patois.
Asking about a sports team.
Questions with ambiguous results If you type “japa” into Google, it brings up a few answers that clearly show that “japa” is slang for something. It’s not the same with ChatGPT. It replied every variation of the word with the default answer of not understanding the question. Even as we tried to progressively ask questions in the hopes of getting the bot to realize what it is, it was a futile attempt. Funny enough, it understood mandem after a few iterations.

Similarly, when I want to catch up with my favorite team, I sometimes search for the said team, and Google knows who they’re playing at the moment, the scoreline, and any other pertinent information at the time. For example, if the coach was sacked the day before and I Google the team today, the sacking would be one of the prominent results. ChatGPT doesn’t have this capability. It shouldn’t. If it did, it wouldn’t be what it is.

Does a user who searches for “Tesla” mean the company or the engineer? Does “rush hour” mean the actual heavy traffic period or the movie with Jackie Chan and Chris Tucker? The list goes on.

Conclusion

Search has its place in our digital world today; a chatbot won’t replace that relevance. Billions of people still look to Google for answers to questions daily. This will remain so for a while. The best part is that we’re glad Sam Altman agrees with our stance on the limitations of ChatGPT.