When Google Starts Talking and ChatGPT Starts Searching

Ever since the time browsers were invented, the world of knowledge became open to us all. Now we didn't have to worry about not knowing things, we had search engines like google to help us. So much so that Google has now become a verb from a noun. We just say Google it! 

I recently read an article on ‘Prompt Engineering and AI institute’ by Sunil Ramlochan, where they discussed OpenAI introducing web search capabilities. It got me thinking—are we witnessing a shift where AI assistants are becoming search engines and search engines are turning into AI assistants? It’s fascinating how the way we find and interact with information is evolving right before our eyes.

A quick history

It's hard to believe now, but just 25 years ago, the idea of "searching" the World Wide Web was practically non-existent for most people. The World Wide Web, as we know it, was only invented in 1989. In the early 90s, it was still a relatively small and obscure network. Most people were not even aware of its existence, let alone how to find information on it. There wasn't a central, user-friendly tool like Google to type in a question and get answers. People who did use the early web relied on word-of-mouth, limited directories, or knew specific website addresses.

For those adventurous few who did venture into doing it, finding information online was still a real challenge. Imagine a library with no card catalog and books scattered everywhere! Early tools like Archie, Veronica, and Jughead were like the first attempts to organize things, but they were very basic. 

Then came the first generation of true search engines. Programs like WebCrawler and Lycos were pioneers in making the web searchable. They emerged in the mid-1990s as some of the first true web search engines, using automated programs to index web pages and allow users to find information based on keywords. These early engines, while primitive by today's standards, played a crucial role in making the web more accessible and useful to the growing number of people online. Then came Yahoo! Yahoo with its human-curated directory, was like a meticulous librarian, carefully cataloging each site.

But the real game-changer was when two students from Stanford University; Larry Page and Sergey Brin came up with the ‘BackRub’. ‘BackRub’ was a research project started in 1996 by these two. It used a novel approach to ranking web pages based on the number and quality of links pointing to them. ‘BackRub’ quickly outgrew its Stanford servers, and in 1998, Larry and Sergey officially launched Google, a search engine that would forever change how we explore the online world.

The Shift

I think that just as the internet brought a transformation in the way we searched for information, there’s yet another revolution! With the advent of large language models on the scene, we are now not satisfied by simply getting search results for our queries, we want more! We want answers to our queries! 

So, while search engines were getting really good at finding things on the internet, something else was happening too. People were trying to teach computers how to understand and answer our questions, just like a person would!

At first, they made simple programs that could chat a little, but these were kind of like toys. Remember those chatbots you used to find on websites, like "Ask Laila"? They followed basic rules and couldn't really understand what you meant.  You'd ask a question, and get a completely random answer, or a canned response that made no sense. It was pretty frustrating! Early chatbots used fixed rules that relied on set patterns, making their responses feel robotic and unnatural. Those early attempts showed that creating a machine that could truly understand and respond to our questions was a much harder challenge than it seemed.

But as computational power grew and algorithms became more sophisticated, a new breed of models called large language models evolved. Trained on massive amounts of text, they learned to recognize patterns and relationships between words. This breakthrough enabled them to generate meaningful responses with greater accuracy—making natural, human-like conversations possible.

The development of deep learning (a discipline within AI) and neural networks marked a turning point. Models like BERT and GPT emerged, capable of understanding complicated things and even writing stories or poems. These were called AI assistants. It was like having a super-smart friend who knows a lot! 

The AI Challenge: Limited, Misguided, Unverifiable

But there's a catch: training these models takes a ton of time and computer power. It's like trying to read every book in a giant library—it takes a while! So, they have to stop feeding them information at some point, which creates a "cut-off date." Anything that happened after that date, the models don't know about yet.

This "knowledge cut-off" was a real roadblock for large language models. They could access a vast amount of information, but it was like having a snapshot of the world frozen in time. What good was knowing everything up to 2021 if you couldn't answer questions about current events, new discoveries, or the latest trends? Ask them about yesterday's news, and they'd say, "Sorry, I don't know anything after my training cutoff date. This was really frustrating when you needed current information about recent events, new products, or breaking news. Plus, these early AI assistants sometimes made up facts because they couldn't check their answers against reliable sources. They also couldn't tell you where they got their information from, so you couldn't verify if what they were saying was actually true. These limitations meant that while they were good conversationalists, they weren't always reliable helpers when you needed accurate, up-to-date information. People loved having conversations with these AI helpers, but they needed up-to-date and reliable information too.

Technologies converge

So what happened next? AI companies came up with a simple solution: give their AI assistants the ability to search the web, just like we do.

This changed everything. Now, AI assistants could:

• Look up current events and recent information

• Double-check their facts against reliable sources

• Show where they got their information from

• Do much more than they could with just their training data

But here's where things get really interesting. At the same time, something different was happening with search engines like Google.

For years, Google would give you a list of links when you searched for something. It was up to you to click those links, read different websites, and piece together your answer. But now, Google has started giving direct answers at the top of search results - kind of like what ChatGPT does.

So we're seeing these two different technologies start to look more and more alike:

• ChatGPT and other AI assistants are becoming more like search engines by looking things up on the internet

• Google and other search engines are becoming more like AI assistants by giving conversational answers

It's like they're using each other's recipes! Or maybe they're just evolving into the same thing: tools that both find information AND explain it to you in a helpful, conversational way.

This is great news for us. We don't have to choose between getting accurate, up-to-date information and having a helpful conversation. Soon, we might not even think about whether we're using a "search engine" or an "AI assistant" - we'll just be getting the answers we need, when we need them. Search engines are evolving to provide direct answers while AI assistants are incorporating search capabilities—converging toward a new hybrid model of information delivery. We're moving from a paradigm of "here's where to look" to "here's what you need to know." and it's such a welcome change! 

So, in the end, does it even matter whether we're "Googling" or "ChatGPT-ing"? Maybe not! What really matters is that getting the right answers is becoming faster, easier, and way more natural - just like having a conversation with a really smart friend.