I probably heard the term “Google killer” 20 years ago. At that time, Google was really coming to the forefront of internet search. Before 2004, search engines such as Ask Jeeves and Yahoo were commonly used, but it wasn’t long before the engine we all know and love (?) usurped them. When new engines launched, it was often with a lot of hype—this time Google would lose out to the new kids on the block—but of course it never happened. Google has remained the predominant force in most of the world, with only a handful of challengers in Russia and China.
Google defined what it meant to search the internet; how search results displayed on the screen, the algorithms to rank results, the use of advertising linked to the search terms used, and some complex search functionality. The latter interested information professionals even as it left everyone else cold. In my opinion, Google reached its peak functionality in about 2012 and it’s been going downhill ever since. It’s always worth remembering that Google isn’t a search engine, it’s an advertising platform that uses search to make money.
And that’s the basic problem. If it’s really good at its job, people run a search, get the information they need by following a link and, while the searcher is happy, Google doesn’t make any money from advertising clicks. On the other hand, if it doesn’t provide helpful results, searchers are more likely to look at adverts (or even click on them inadvertently), which does make money for Alphabet. Thus, it has to walk a difficult tightrope—be good enough so that people keep coming back, but not so good that it loses out on advertising revenue and not so bad that people look elsewhere for their information. To be fair, Google’s done a pretty good job over the last two decades.
People want to find, not search
However, its success is also its downfall. The results screen is now littered with adverts, snippets, news and so on. It’s complex, messy and unattractive. At the heart of the problem is that people don’t want to search, they want to find. The seemingly endless display of link after link is not helpful, as the search engine becomes the gatekeeper to the information searchers want. They have to go through page after page, clicking on links in the desperate hope they’ll get the information they need almost by serendipity. If it’s a particularly complex problem, it’s going to be necessary to click on several links, gathering pieces of information along the way before being able to get a nice rounded-out answer. We all know this and I’m not telling you anything new. It was what it was, and we all accepted that was how search worked.
In December 2022, most of the world woke to find ChatGPT had exploded onto the scene, with 5 million users inside a week, and currently at 100 million regular users and 1.6 billion views a month. In those early days, there was much discussion about how useful ChatGPT actually was—the database it used was elderly and prone to confidently hallucinate its answers. Nonetheless, it did prove a useful tool—I would encourage clients to use ChatGPT to suggest search terms and phrases that could be used in the search boxes of traditional engines.
To be fair, it was pretty good at doing that. It was also useful to read around a subject prior to going and doing proper ‘serious’ searches. But a search engine it certainly wasn’t. What it was good for, though, was to give a nice overview of a subject—one of my very first searches was “What were the causes of the American Civil War?” and I was stunned at how helpful and interesting the answer was to that very basic prompt. I thought to myself, “This is what people want—they don’t want to search, they want to find.” Of course, there were problems—the out of date database it used and the lack of transparency as to the sources from which it drew its results.
I could look at the answer the chatbot gave and it made sense. Yes, it got things wrong now and then, but overall I was impressed. What wasn’t evident then, and still isn’t evident to a lot of users now, is that ChatGPT doesn’t know anything, nor do any other GPTs. It looks as though it does, and what it says generally makes perfect sense, but it’s not a knowledge base cherry-picking information from one place and adding that to content it got from elsewhere.
It is simply a large language model that is an extremely advanced statistics engine. There is no agent behind the scenes pulling it all together, but instead it’s a large, very advanced statistics model trained on our language. So no, ChatGPT isn’t a search engine in its own right, but what it did do, right from the start, was show us that there is another way of answering queries. Instead of screen after screen of results, we were presented with an answer. It might not always have been correct, but it was a welcome change. It wasn’t necessary to search because the answer appeared on the screen in less time that it took me to type out this sentence.
This is a fundamental change in the way that search works, and that cannot be stressed enough. I’m old enough to remember the impact that CD-ROM technology had. It meant that people could start to do their own searches from their desktops and it made information more available quickly and easily. It spawned a whole industry as a result. This change is far greater and will have a more profound impact.
The situation today
The current situation is in a state of flux, which is reminiscent of Google changing the early internet search industry, and later of Web2 morphing into social media. So let’s take a look at what is happening now. We are seeing a new breed of search engine. Perplexity, You.com, Gemini, Copilot, Consensus and several other search engines exploded onto the scene in the last year. Their approach to search is similar to the results you get from a search with ChatGPT, only better.
All of these new engines provide a packaged answer to the prompt or question. We have finally reached the point searchers have long wanted—instead of screens of webpage results, we have a nice neatly packaged answer. Moreover, the results also include sources used, so it’s possible to actually go and check the webpages themselves. Indeed, using Copilot, users can check the page itself, get a summary of the content, and ask questions about it, via the flyout screen. Alternatively, they can take the document, paste it into a Chatbot and use that to reformat the content or ask more questions.
Users can also carry on the conversation, and this is another game changer. In ‘the “old days” (last year), you would have to restate your question when using Google, because it didn’t remember what your last query was about. That’s no longer necessary, since the new tools remember your search history. You can ask more questions without having to explain to the engine what you were looking for or put it into context. Many of these engines will also provide context-specific follow up questions as well. Search has transformed into conversations, leading to better answers, and giving the searcher a far richer response than has been possible before.
Returning to ChatGPT, you can now create custom GPTs that will give better and more accurate results based on the resources you give them. The new version ChatGPT-4o (the o stands for omni) is light years ahead of even the paid version. Using my mobile phone I can talk to ChatGPT and it will talk back to me. It will be aware of my surroundings via the camera, which impacts its answers to my prompt. We are almost at the point where we can dismiss text completely and simply have a conversation with an avatar that is almost indistinguishable from a real human being. Search has come off the page and into everyday life, quite literally.
Traditional search is dead
Where does that leave traditional search? I’m very confident in saying that it’s dead. It may not quite realise it yet, but that’s the truth of the matter. I began this article by mentioning “Google killers”. Google isn’t going to die—and the Gemini product makes it quite clear that the 800-pound gorilla isn’t going away. But what AI search has killed is the traditional approach that proved so compelling in the past. All of the older engines are going to have to adapt remarkably quickly and incorporate AI into their offerings. Microsoft is clearly way ahead of the game here with Copilot and it’s become my preferred search engine because I get the answer that I want, not some halfway house of a list of pages to go off and look at.
We don’t have a Google killer, what we have is a Google transformer. Where this leaves us with regards to advertising I’m not entirely sure. I would imagine that Google will certainly want to drop adverts onto our screens, either before or after the packaged result (or knowing Google propensity to throw adverts at us at every opportunity—both!) but is anyone really going to be interested enough to look, when they’ve already got the answer to the query?
Yes, we will still need to search for local restaurants and so on, as we have always done. However, here’s my vision of the future: You’ll be out and about in a strange city; you’ll pull out your AI assistant, which will look through your camera and work out where exactly you are, the time of day, and what the weather is like. You’ll then simply ask “Where’s a good place to eat?” and your assistant will be able to base its answer on what it has previously learned about your preferences and search its knowledge base about where local restaurants are. It can then direct you to the restaurant. Even more—and I don’t think this is too much of a flight of fantasy—it could see that it’s raining and that you don’t have a coat and find somewhere close so you won’t get too wet walking there.
Where does this leave information professionals who want more from AI tools than restaurant recommendations? If people can do all of that for themselves, does that make us redundant? I don’t think it will. In fact, I think that AI is going to be one of the best things that’s happened to us since the invention of the internet. At the very basic level, it is still going to be all about information. Asking for it, assessing it, and adapting and reusing it. That’s what we do best. We can start to take the information that we get, repackage and repurpose it. We can take a PDF and feed it through a chatbot, get the information back in a different format then take that and give it to an AI tool that will create a slideshow. Or a video. Or a podcast. Or a blog. Or anything that we want.
We are strides ahead of everyone else because we know how to ask the right question in the right way and then decide how best to present it to our enquirer. But we can only do that if we understand AI—not just search, but also the myriad of other tools that are available. Crucially, because we know information, we are in the best possible position to expand on our skillset and produce new content and services that until now hasn’t even been a pipe dream. We positioned ourselves as leaders when the internet first came out; we adapted our approaches with the development of new search engines; we embraced the use of social media; and we are going to excel when it comes to AI.
Phil Bradley is a U.K.-based internet search trainer and consultant, currently offering online Power Hour courses.