Reading Time: 27 minutes -

Running a digital marketing agency, especially one that specialises in SEO, means confronting some important questions about the future of SEO in a world that’s now consumed with AI and tools like ChatGPT. Without a doubt, the future of SEO is set to change dramatically. We’ve previously written about the AI Inflection Point created by ChatGPT, but the true impact of this on search engine optimisation is still unknown.

Seeing headlines claiming the “death of SEO” is nothing new – we’ve seen it countless times before. In fact, this report dated November 2020 states that:

“According to Ahrefs, SEO has died 4,852 times since January 2016”

This was two years before the release of ChatGPT. A simple Google search will also show around 90,000 to 100,000 results indexed for the terms “SEO is dead” and “Death of SEO”, with around 3,500 of these within the past year alone.

Understanding this dilemma is very important to me as a business owner but I expect it is for anybody working within the industry or reliant upon digital marketing, so I’ve spent a great deal of time educating myself about AI and trying to answer the following questions:

• Can SEO be killed?
• Will AI replace SEO and ultimately search engines?
• What does the future of SEO look like?

These are core themes that I will explore today.

Some might wonder whether this even matters to the average business, or is it just SEOs and perhaps the search engines themselves, that should be concerned by the threat of AI and solutions like ChatGPT. The reality is that all of us are reliant on search engines to some extent and everybody should be concerned enough to consider what AI means to them.

“In August 2023, a survey conducted by research company GlobalData found that more than 50% of businesses questioned had experienced significant disruption within their sector as a result of AI.”

The above comes from verdict.co.uk highlighting the situation we’re all in and the impact it can have on our livelihoods. It’s a sad fact that AI WILL inevitably put many individuals and companies out of business. For others, however, AI is an opportunity – and perhaps now is the ideal time to pause, breathe, reflect – and potentially evolve.

So, without further ado, let’s take that moment right now, and consider the future of SEO, the potential threat or opportunity posed by AI, and the impact this will have on online search as we know it.

A Not-So-Short Intro

The Rise of LLMs and ChatGPT as a Potential Rival to Search

As a technology lover and a marketer, I’ve never been as impressed or as excited as I’ve been this last year experimenting with ChatGPT and similar LLMs (Large Language Models) including Google’s Bard and Anthropic’s Claude 2 with its impressive 100K context window.

Sad, I know, but the opportunities really do seem endless.

For paid (or Plus users), ChatGPT alone is an incredible tool that can do so many amazing things that would have seemed impossible a year or two back.

I’m not going to cover how well ChatGPT does in this article, but here are some of the current capabilities:

• Answer questions in a style that’s no different to having a text chat with a friend
• Write entire pages of text with 100% accuracy in terms of spelling and grammar
• Analyse thousands of words of text and summarise, evaluate or critique its content
• Browse the web in real-time using Bing or other browser plugins to query the results
• Analyse images that are uploaded and view them as a human would, able to understand the context and answer questions
• Create original images using Dall-E 3 using simple text queries (prompts) – these images can even create text within the image now
• Process multiple uploaded documents (MS Word, PDFs, Excel spreadsheet, etc) and answer questions about their content
• Convert from one file format to another, e.g. from a single spreadsheet to many individual Word documents
• Create, review, explain, and even execute code
• Review CSV data and even raw unstructured data to create charts, provide detailed analysis, and produce reports

I’m not even scratching the surface here, but you may be wondering what all this has to do with the future of SEO or search engines.

Microsoft’s Bing Chat and Partnership with OpenAI

Well, for starters, in 2020 Microsoft partnered with OpenAI, the company behind ChatGPT to become the first player in the race to become what many are now terming an AI-first search engine. We’ve quickly seen Bing Chat offer a service very similar to ChatGPT but with the added bonus of being able to query Bing search results out-of-the-box.

Bing Image Creator has become an alternative to OpenAI’s Dall-E product, able to produce high-quality AI images within 60 seconds or less.

The Bing Chat feature is now integrated as part of Bing’s traditional search engine providing users with the best of both worlds – traditional search and AI search.

Google’s Bard and AI Search Initiatives

Google has been later to the game in terms of this “new style” of generative AI, but they’re no stranger to AI.

Google’s most well-known venture into this space was in 2015 with its RankBrain algorithm but the company has otherwise been relatively quiet about AI until ChatGPT was launched.

Shortly after Microsoft launched Bing’s AI capabilities, Google came out with its own version called Bard and we started to hear about other initiatives, such as LaMDA (Language Model for Dialogue Applications), which has actually been around for some time. There was also PaLM-2 (Pre-training with Abstracted Language Modeling-2) and the Magi project.

Since May 2023, Google has been discussing the radical overhaul of its traditional search engine to incorporate AI search.

What About SEO?

Clearly, the search giants are taking note, but what about everyday individuals and businesses that rely on SEO?

Well, the first and most obvious thing, is that tools like ChatGPT can analyse, proofread, and write vast amounts of content, which on the surface looks as good as any human can write. This has already put SEO copywriters out of business, with many businesses now turning to tools like ChatGPT or Bard to produce AI-generated instead of employing or paying a copywriter.

SEO is so much more than just copywriting though and involves aspects such as user intent analysis, keyword research, analytics, technical optimisation, outreach, marketing, and so much more. You guessed it, ChatGPT can automate much of this as well.

So, at this stage, you might be thinking that the death of SEO is inevitable, either because it won’t be needed any longer with how search engines are changing, or because tools like ChatGPT will make manual SEO tasks and SEO professionals obsolete. We’ll revisit this important question later.

The Evolution of SEO

Keywords and Backlinks

When I first got involved with SEO in 2009, it seemed like a dark art and minefield to understand, but I soon realised it mainly boiled down to two main ingredients:

• Identifying the right keywords (and meta tags)
• Building links back to the website

This worked an absolute charm until fast-forward a few years to 2011, and suddenly Google’s algorithm updates like Panda and later Penguin were turning the SEO world upside down. If you weren’t adapting, you were basically signing your site’s death warrant. I recall many in the industry touting the death of SEO even back then.

And yes, the SEO back in the day isn’t the same SEO we all understand today, and today’s SEO won’t be what we understand tomorrow.

People-First Content

Today’s SEO isn’t just about optimising for search engines anymore it’s about optimising for the user.

This was made even more transparent with Google’s Helpful Content Update (HCU) clearly states:

“We recommend that you focus on creating people-first content to be successful with Google Search, rather than search engine-first content made primarily to gain search engine rankings.”

Factors such as user intent, user experience, mobile optimisation, and website speed have become critical. Concepts like E-A-T (Expertise, Authoritativeness, Trustworthiness), or more recently E-E-A-T (Experience, Expertise, Authoritativeness, and Trustworthiness), have become pivotal in Google’s evaluation of websites.

The search engines have moved from being a “dumb” catalogue of web pages that can be easily manipulated through black hat SEO techniques such as keyword stuffing or link spamming, to sophisticated systems. Ever since the introduction of RankBrain in 2015, AI has played a role in sorting search results. These days, AI models like BERT are helping Google better understand the context and semantics of a query.

In fact, the more somebody tries the “game the system” or produces search engine-first content, it’s highly likely to backfire.

This is one of the reasons why businesses turning to ChatGPT or other LLMs like Bard for SEO and content creation will most likely end up getting hurt without the knowledge and expertise of an SEO who’s been in the game.

From SEO to Algorithm Optimisation (AO)

A relatively new phrase has been termed ‘Algorithm Optimisation’ or AO.

From an SEOs perspective, this almost sounds the polar opposite of a people-first strategy, but this is actually about crafting content that resonates with the platform’s specific algorithm to boost visibility and engagement.

While people-first is a good principle to follow, it’s also essential to understand the platform and its algorithms, as well as the end-user.

For example, on Instagram, it’s not just the quality of your photo that matters, but also how quickly people engage with it. On LinkedIn, long-form content that gets people talking might score you more points with their algorithm. On YouTube, videos with a higher number of views, comments and likes are likely to rank higher than those without.

In essence, AO on social platforms is quite different to SEO on search engines. It means adapting your content strategy to what each algorithm values most, be it user interaction, relevance, or timeliness, to ensure you can get found and reach a broader audience.

“Death of SEO”

About every couple of years, someone will shout from the digital rooftops that SEO is dying, thanks to the ‘next big thing.’

Be it penguins or pandas, social media, or AI, the doomsayers will always predict the end of SEO. Yet here we are, still discussing the future of SEO. Why? Because the need for search and the role of SEO keeps evolving.

The way people search might evolve, be it through a search engine or an AI chat interface, but the need to find accurate information quickly isn’t a fad, it’s a permanent fixture in our lives.

When thinking about the style of SEO that worked back in 2009, the main change is that the once siloed discipline now needs to be a more integrated and holistic part of the overall digital marketing strategy. Putting people first is obviously the right approach, but that also means understanding a complex array of factors including user behaviour, market sentiment, and competitive positioning.

The reality is that SEO is not dead but it will continue to evolve like it always has done. Whatever name it goes by (SEO, AO or AIO), understanding people, data, algorithms, and platforms, will continue to be important.

AI vs SEO & Traditional Search

The Similarities

As I began to experiment with ChatGPT, and later Bing Chat, Bard and Claude, it struck me that both traditional search engines and these chat interfaces aim to do one thing: answer a user’s query as accurately as possible.

The obvious difference is the ability to have a human-like dialogue with these AI chatbots and get remarkably good answers.

Both sift through heaps of data to provide the most relevant answer.

And increasingly, both understand not just the input, whether that be short keyword searches, a longer tail search query, or a prompt, but they also understand the intent behind them.

The Differences

Despite the similarities, there are more differences.

User Interface: 10 Answers vs 1 Answer

A search engine serves up a list of links, typically the 10 blue links we’ve become conditioned to expect on the first page, potentially with a Local Pack, People Also Ask, and various images, videos and shopping results thrown in. If we’re not happy with the first 10 results, we move on to the next page.

As anybody involved with SEO will know, it’s all about page 1, ideally the first few results. Users only tend to go to page 2 or further in extreme circumstances.

Using chat interfaces like ChatGPT, Bing Chat or Bard, we get a more conversational experience. It feels less like querying a database and more like asking a well-informed friend. The way the information is presented is simplistic, generally with an input box for the prompt and a results box for the answer.

In some ways, this isn’t so different from a search engine, but instead of getting 10 blue links, we now see a single response that directly addresses the user’s input (prompt). Although, the user can always ask for more than 1 response. I frequently ask for alternatives and options when using ChatGPT so I can compare.

Links and Traffic vs No Links and No Traffic 

The other major difference is the absence of links.

Search engines by their very nature are designed to provide a list of results where users can click through to the website.

ChatGPT by default won’t include hyperlinks or provide any sources and citations for the answers provided. This means verifying information is a challenge but it also means that the sources and owners of the underlying information won’t get attributed or awarded with traffic.

However, paid ChatGPT Plus users can use ‘Browse with Bing’ or a browser plugin such as BrowserPilot or WebPIlot with GPT-4 to access the internet or a web page. Even then, the results don’t tend to include links and are unlikely to generate any traffic back to the source.

Up-to-Date Knowledge

ChatGPT’s knowledge only goes as far as September 2021, meaning it has no awareness of recent or current affairs as standard. The obvious benefit of a traditional search engine or the ability to use ChatGPT Plus (‘Browse with Bing’ or a browser plugin) is that we can access real-time (or close to real-time) information. More on this later.

Hallucinations

LLMs like ChatGPT are known to make up facts and information. In some cases blatantly lie, giving false information, which is commonly referred to as a hallucination. These hallucinations can seem so convincing that many users simply believe them to be true.

This was more common with the earlier GPT-3 models than we see using GPT-4, but this is something a traditional search engine is incapable of doing.

That said, information within a search engine index can just as easily be untrue or out of date. Whether it’s an AI chatbot or a search engine, the onus is always on the user to verify and fact-check the information given.

ChatGPT vs Google For Search: Understanding The Mechanics

Given the similarities, it’s not surprising that many ChatGPT users expect it to work in a similar way to Google.

We add an input and get a response, so the overall mechanics must be very similar, right? Wrong.

When I first started using ChatGPT, I kept hearing the term ‘knowledge base’ being used. In my SEO mind, this was the equivalent of Google’s search indexed, and ChatGPT was simply using some kind of AI magic to query an out-of-date version of Google’s (or perhaps Bing’s or various others) index.

The reality is that they are very different.

Google Search

Think of Google Search as the librarian of the internet but on steroids.

When you type a query into Google, it doesn’t just look at the surface; it digs deep into its index, essentially a database of indexed websites, to find the most relevant results. The Google algorithm takes multiple factors into account, such as keyword relevance, website authority, user experience, and much more.

Google uses complex algorithms, some of which we’ve discussed above. These algorithms are kind of like smart little workers (middlemen) that query Google’s index and decide what results to return to the user. Along with its index, these algorithms are constantly updated to try and serve you the most relevant content.

ChatGPT

ChatGPT is built on a powerful neural network model.

This architecture allows it to understand context, generate human-like responses, and even engage in creative tasks.

It does this by drawing from a massive dataset, which is analysed using machine learning techniques. So, it’s not just answering queries – it’s making calculated responses based on data patterns it has recognised.

Many people (like me) probably start off by thinking LLMs like ChatGPT query a database or index similar to Google. This just isn’t the case.

To try and understand how it all worked I started by reading Stephen Wolfram’s article here. It’s a long read so I asked ChatGPT to summarise in a story-style form (here) so that children as well as adults could understand. This part in particular stood out for me:

“ChatGPT is like a super smart friend who can guess what words should come next in a sentence. It creates stories by putting words together in a way that makes sense. It’s like magic!”

For a slightly more detailed version, ChatGPT is essentially trained using massive amounts of text data and is taught to anticipate subsequent word sequences in a given situation. The text generation is driven by the patterns they’ve picked up while being trained, rather than scouring the internet or indexing web pages.

Mechanics Comparison: Search Engines vs. LLMs

Steps/Processes	Search Engines (Google)	LLMs (Language Models)
Data Collection	Crawls the web to collect data from millions of web pages. Analogy: A spider crawling over a web.	Gathers training data from various sources like books, articles, and websites. Analogy: Assembling ingredients for a recipe.
Data Organisation & Preprocessing	Organises data into an index stored across multiple servers. Cleans and filters data. Analogy: Sorting and preparing ingredients before cooking.	Cleans, filters, and processes raw data to remove irrelevant or sensitive information. Analogy: Cleaning and preparing vegetables before cooking.
Processing & Training	Uses algorithms to sift through the index and find relevant results when a query is entered. Analogy: A chef selecting the best ingredients for a dish.	Trains the model using processed data to learn patterns, grammar, facts, and reasoning abilities. Analogy: Baking a cake with the right mix of ingredients.
Knowledge & Relevance	Uses signals like keywords, backlinks, and user behaviour to determine the relevance of a page. Analogy: A librarian recommending books based on popularity.	Doesn’t “know” facts but recalls patterns when queried. Has seen a vast amount of information during training. Analogy: A librarian fetching relevant books based on your question.
Continuous Learning & Feedback	Updates algorithms based on user interactions, search trends, and feedback. Analogy: A gardener pruning and fertilising plants.	Continually refined based on user interactions and feedback. Analogy: A sculptor constantly chiselling and refining a sculpture.
Technical Infrastructure	Supported by a robust infrastructure of servers, data centres, and complex algorithms. Analogy: The engine room of a massive ship.	Runs on powerful computational infrastructure, including GPUs and TPUs. Analogy: The engine and machinery behind a high-speed train.

While the end goals of search engines and LLMs may differ, the underlying processes share striking similarities. Both systems are complex, data-driven, and rely on robust technical infrastructure.

So, there you have it: two distinct technologies – LLMs like ChatGPT and search engines like Google – each with their unique mechanisms but undeniably similar processes for helping us find and understand information.

But, Can LLMs like ChatGPT Replace Traditional Search? The Challenge of Real-Time Data

The Importance of Real-time Data

As impressive as the answers are from LLMs like ChatGPT, they are not based on up-to-date knowledge. Of all the differences between traditional search engines like Google and LLMs like ChatGPT, this one will ultimately determine whether the likes of ChatGPT will replace Google.

You might now be thinking, you said ‘Browse with Bing’ or a browser plugin can access real-time information so what’s the problem? This is true, but that information is dependent on external sources such as search engines with their index to get the latest information.

By doing this, LLMs continue to be reliant on search engines. Not only does this mean we are getting information from two different places, but it also means that search engines, and by necessity SEO, won’t be disappearing any time soon.

How Traditional Search Engines Handle Real-time Data

Search engines like Google and Bing use web crawlers, often known as bots or spiders, to scour the internet and index web pages.

These crawlers seek out new content and look for updates to existing pages, adding them to the search engine’s index. However, this process isn’t instantaneous. Depending on various different factors, it can take some time for new or updated information to appear in search results.

This is where SEO comes in. One of the most basic and fundamental aspects of SEO is to ensure a website can be found and indexed easily. This includes techniques, such as:

• ensuring the website has a clear navigation structure and search engine-friendly URLs
• including breadcrumbs to show the relationships between pages
• adding a sitemap (HTML and XML)
• including the correct meta tags to allow indexing and crawling from one page to the next
• setting up the likes of Google Seach Console and Bing Webmaster Tools and submitting the sitemaps
• having a robots.txt file linking to the sitemap

All of the above will help ensure that a website and its pages can be found.

Other factors will determine how often a website is crawled and added to a search engine’s index, such as the age of the website, the authority of the website, and how often it’s updated. On the whole, these are also aspects we can influence using SEO.

So, when it comes to real-time data, traditional search engines have a bit of a lag, but they are currently the best place to go for a broad range of up-to-date information. Other sources like Twitter and Reddit are often the preferred choice for the very latest real-time information – and providing these are public profiles, ChatGPT can browse them too using plugins.

The Technical Nature of ChatGPT

ChatGPT, unlike traditional search engines, doesn’t crawl the web in real-time to fetch new data. Its training dataset is frozen at a specific point in time.

By default, this means that LLMs like ChatGPT will never be able to replace search engines or SEO.

The most likely scenario for them right now is that they get integrated into traditional search engines in exactly the same way as Bing has done and Google is in the process of doing right now. Keep reading to understand more about the latest evolution of search and how it is becoming more intertwined with LLMs and generative AI.

The Current Evolution of Search

AI-First Search Engines

In February 2023, Bing announced “The New Bing” aka the new AI-powered Bing search engine (Bing Chat).

In February 2023, Google announced Bard, but not long after this, in May 2023, Google announced its generative AI search, giving a preview of its AI search engine.

In both examples, we see a hybrid of the traditional search engine we are used to but integrated with ChatGPT-style capabilities and generative AI results.

We have the ability to ask follow-up questions, see shopping results based on the results of previous dialogues, and have media presented in a variety of different formats.

For somebody like me who’s a lover of tech, AI and SEO, this really is quite exciting! Again this doesn’t indicate the death of SEO, only that the future of SEO will be very different and we’ll all need to adapt our strategies within this new AI world.

What Will Change and What Will Stay the Same? 

Search engines will still need an index and that index will still consist of websites, so no major change here.

Website owners will still want to be found, and at the other end of the spectrum, users will still want to find the most relevant results as quickly as possible. Again no major change from the situation right now.

Both of these factors mean SEO will be required, so for those of us who try to keep up, we can be confident that we won’t be put out of work.

However, there will be big changes ahead and factors we need to consider, including:

• Search engine algorithms will presumably continue to provide traditional search results, alongside or even integrated with generative AI results.
• The generative AI results will be different for each search engine, in exactly the same way that traditional search results are different in Google and Bing, or Bing Chat provides different answers to Bard. Bing’s LLM is based on ChatGPT and Google will have its own LLM called PALM-2 (no wait, the latest reports seem to suggest it will be an LLM called Gemini), to compete with OpenAI and Microsoft. It tends to get confusing with all these acronyms bouncing around.
• The way search engines decide to integrate the generative AI results will be different. Currently, Bing Chat is a small icon next to the search bar, meaning users can easily toggle between traditional search and Bing Chat (or AI search). Google’s demonstration seems to suggest a seamless integration of generative AI results mixed in with traditional search results, meaning users may not have a choice.

For those dependent on these decisions, i.e. the website owners, end users, SEOs, and marketers, this will almost certainly change, and perhaps not for the better.

I think we can fairly confidently say that these new AI-first search engines will do a very good job of meeting the needs of the end user and they will be happy.

Google Search Integrated with AI: An Overview

Trying to picture how this new AI-first Google search engine would work, I imagine something like this:

• Google’s native search results would serve as one data source for the AI/LLM.
• These results would be queried by the AI to provide answers based on Google’s real-time index.
• Additionally, the AI/LLM would use its own dataset (training data) as a second source for generating responses.
• The user would see both AI-generated answers and traditional search results.

I attempted to draw my own picture of how I see this working but it failed dismally so I asked my best friend ChatGPT to draw some pictures using the “show me” plugin.

We came up with three different illustrations, so depending on how you process information, hopefully, one of these will make sense.

The first one was a mindmap. This shows the various components but seems a bit overly simplistic for my liking.

The next one shows a sequence diagram. This one makes more sense to me in terms of the various relationships and how both traditional search and AI search might work together.

Finally, my personal favourite below shows all the various interactions and how Google’s AI would query its training data but also the web index alongside the algorithm for traditional search to return the combined search results.

It’s possible these could all be wrong, and of course, the pictures showing how “The New Bing” works would look a bit different.

What Challenges Does This Pose?

Taking a few steps back from the technical stuff, let’s think about what challenges this poses for the website owners who want to be found. For example:

• With AI-generated results taking centre stage, will the standard ten blue links (organic search results), which have already been demoted below sponsored results, local results, ‘people also ask’ results, and so on, be pushed out of existence?
• Will traditional search results be replaced altogether with results provided by the AI that simply reference the source?
• What will happen to paid/sponsored results and where will they appear?
• Will the AI-generated results be able to integrate both organic and paid results into their answers? Would this even be fair?

There are still so many questions regarding the future of search engines and what this means for the individuals, businesses, and organisations reliant on them. All of this would also impact the SEOs and marketers whose job it is to help businesses to get found.

The reality is that we can all expect a rate of change during the coming years like no other time in recent history. We’re all finding our way in this new AI world, and that includes the search engine giants and the brains behind them, meaning what’s presented today as a solution might be very different tomorrow.

Make or Break Factors for AI-First Search

The conversation surrounding AI-first search engines involves several core elements: the trust users place in AI-generated responses, potential avenues for search engines to monetise this new style of search, and the user experience.

So, let’s unpack these considerations.

The Trustworthiness of AI-Generated Responses

Trust is like a currency in the online space. For search engines to keep users coming back, there has to be a level of trust between the user and the algorithm.

Historically, search engines like Google have built this trust by striving to provide the most accurate and relevant results.

But as we move into the world of AI-generated responses, questions about trustworthiness inevitably arise. If the AI-generated answers stem from a mixture of the LLMs dataset and the results returned from the search engine’s index, we’re trusting that the AI will give us the right answer and there will be no hallucinations.

Search engines would need to tone down the AI’s “creativeness” to avoid hallucinations and help ensure that we get the right answer.

Any lapse in this could result in AI responses that are less reliable or even misleading.

Right now, users need to trust that Google’s algorithms are working and providing the best results. Users then have a choice to select which results they want to view, usually starting with the first blue link ranking in position 1, then moving down the list based on whether the search results match what they are looking for.

In other words, the user is empowered to select the results they like.

With AI-first search, that empowerment may be taken away, or at least partially, with the AI result taking centre stage.

Potential Monetisation Strategies

Search engines, traditionally, make money from their advertising platforms like Google Ads and Bing Ads. But the integration of AI into search raises some tricky questions about how this monetisation model might adapt and how they will balance impartiality with their commercial interests.

If AI-generated responses include paid or sponsored content, users may feel manipulated and lose trust in the platform. The flip side is if there are no ads, how does the search engine sustain itself financially?

One of the charms of ChatGPT lies in its seeming independence from commercial interests, presenting a straightforward, ad-free interface, with bells and whistles for paid ChatGPT Plus users. It’s a nice model as both free and paid options remain impartial.

Google on the other hand has an obligation to its customers who have historically relied on its advertising services.

It’s a fine line to walk.

One possible solution could be a tiered system where basic AI interactions remain free on Google but supported by ads, while premium could allow for an ad-free experience that would be available for a fee. This would allow for a diversified income stream without necessarily compromising the impartiality of the AI for those willing to pay.

I’m not a fan of this idea as any kind of paid or sponsored input into the AI-generated results would make them biased and lose my trust.

Google’s demo seems to leave this important question out.

I imagine we’ll end up with something similar to what we have now with a mixture of sponsored results, AI-generated results (free of ad input), and traditional (organic) results.

The other alternative is that we might see AI-generated results (free of ad input) with the follow-up button, source link, and possibly related sponsored ads below.

The Impact of AI on the User Experience

AI is impressive when it comes to understanding context and offering nuanced responses, which can be a big win for user experience.

However, the trade-off could be an erosion of the traditional “ten blue links” that users are accustomed to, potentially making the search experience jarring for some.

For instance, Google’s approach appears to be a seamless integration of AI-generated responses with traditional search results. This could be disorienting for users who prefer the existing layout and want the option to choose between AI and traditional search.

I wonder whether we’ll end up with the AI-first version of Google presented to users as the default search engine with a “prefer the old look” option for those who don’t or simply prefer not to use AI.

What’s clear is that the adoption of AI in search is not just a technological shift, it’s an ecosystem change which will have a wide-ranging impact on everybody.

Will ChatGPT Replace Google?

This question is potentially more along the doomsday vibe for SEO and those who claim the death of SEO, and it all comes back to the challenge of real-time data.

If we imagine a world where ChatGPT and its training data are not only up-to-date but the responses are accurate and void of hallucinations, what would be the need for a search engine, its index, or SEO for that matter? This is something I’ve given some consideration to, in terms of whether it’s possible and whether it would make sense.

For example, in such a scenario, there would be questions such as:

• How would the dataset get updated?
• Who would check and validate the new data, and how?
• How would the data be added?
  • Would LLMs crawl the web looking for updates, similar to a search engine?
  • Would users submit their websites or data to the LLM for consideration (more like a directory)?
  • Would an intermediate (a form of moderator) be required?
• How quickly could the LLM be trained on any new data?

Again, there are lots of questions, but let’s look at some possibilities for how this could work.

Collaboration with Traditional Search Engines (The Current Scenario)

How it Works: ChatGPT could integrate with other search engines to pull real-time data, which is exactly how it works today with Bing Chat. The alternative which is in place today is that ‘Browse with Bing’ and plugins allow ChatGPT natively to access search results and data from external sources.

Feasibility: This is the most feasible and straightforward solution. It doesn’t require ChatGPT to have its own real-time index but leverages existing technology. It also means that ChatGPT will never replace Google.

User-generated Real-time Updates (Possible But Unlikely)

How it Works: Imagine a directory or Wikipedia-like model where users can update ChatGPT’s knowledge base. This would require a robust verification system to maintain accuracy.

Feasibility: While intriguing, this model presents challenges in data verification and could be resource-intensive.

Hybrid Model with Periodic Updates (Also Unlikely)

How it Works: ChatGPT could maintain its static knowledge base for evergreen content while pulling in real-time data for specific queries.

Feasibility: This could be a balanced approach but would require complex algorithms to determine when real-time data is necessary. It would also mean that ChatGPT would never replace Google due to its reliance on an external search index.

AI Self-updating Mechanism (Even More Unlikely)

How it Works: ChatGPT could be designed to crawl the web itself, updating its own knowledge base in the same way that search engines do today.

Feasibility: This is highly unlikely due to the enormous resources required to train AI models to handle this level of complexity. It would also require ChatGPT to move away from its core functionality of language prediction to take on tasks like data storage and retrieval, which are entirely different ball games.

An AGI ChatGPT (This One Gets Our Vote)

AGI is short for Artificial General Intelligence. It’s essentially what we all think AI is, i.e. a super-intelligent self-learning robot that could potentially replace humans.

Right now with the current AI and LLMs that we are aware of, this isn’t possible. We’ve actually covered AI vs AGI and the possibility of GPT-5 being AGI in depth.

As the article explains, some believe OpenAI have already achieved AGI internally and believe the next model, GPT-5, could amount to an AGI ChatGPT.

If so, this would potentially address the limitations of current LLMs like ChatGPT or GPT-4 when it comes to real-time data. Here’s how:

Real-Time Learning and Adaptation: Unlike current LLMs, an AGI ChatGPT would have the ability to learn and adapt in real-time. This means it could potentially crawl the web, index data, and update its knowledge base continuously, thereby providing real-time information.

Contextual Understanding: AGI ChatGPT would have a far superior understanding of context, allowing it to not only pull data but also understand the nuances and implications of that data. This would make its responses more accurate and relevant.

Decision-Making Abilities: One of the most significant limitations of current LLMs is the lack of decision-making abilities. An AGI ChatGPT could make reasoned judgments based on the data it has, which would be continuously updated.

Multi-Domain Expertise: AGI ChatGPT would have the ability to understand, learn, and apply knowledge across different domains. This means it could provide expert-level responses in various fields, from medicine to law to technology, all backed by real-time data.

If an AGI version of GPT-5 or AGI ChatGPT were to come into existence, it could be a game-changer for SEO and search engines.

The lines between what a search engine does and what an AGI could do would start to blur. SEO strategies would have to adapt considerably to a landscape where real-time, highly accurate information and insights are readily available from AGI-powered platforms.

My money is on this scenario for a number of reasons. Ultimately, this could mean that ChatGPT would replace Google as there would simply be no need for a traditional search engine or SEO, which I’ll come on to shortly.

The Future of SEO: Generative AI SEO (GAISEO)?

The nature of search engines is inherently about “pull marketing” – users express what they want and expect the search engine to deliver. SEO facilitates this process by helping websites to rank in search results and get seen by the user.

ChatGPT somewhat fulfils this by offering precise and context-specific responses to user queries. However, it doesn’t give a list of options like traditional search engines, which allows users to pick and choose. It also doesn’t rely (or at least not directly) on SEO. So, while it’s aligned with pull marketing, it takes a slightly different approach.

Does SEO influence ChatGPT?

I said “not directly” above because ChatGPT’s training data is made up of a collection of external sources, including websites.

One of the first things I did when I got access to ChatGPT was test whether it knew about me and Opace. Again, sad I know. Unsurprisingly it knew nothing about me but did have a good idea about Opace and what we do.

This leads me to believe that SEO has some role in this, i.e. the more discoverable content is, the better chance it has to be included within ChatGPT’s training data.

Will AI Lead to The Death of SEO? 

While some believe that AI will render SEO obsolete, the human touch, insight, expertise, and trust can’t be replicated by any AI model.

While algorithms and AI solutions can process data and give us answers, they can’t (currently) understand user intent or emotions the way a human can.

Regardless of the changes ahead, the endgame remains the same, i.e. delivering the most accurate and relevant information to the user. Therefore, the principles of SEO – search intent, keyword research, discoverability, and quality content, will continue to be important, albeit with some adjustments to accommodate the new tech on the block.

The key takeaway being re-iterated here is that we can expect a form of SEO evolution, but its future is still strong.

An Omnichannel Approach: Is This The Way Forward? 

The concept of Algorithm Optimisation (AO) was covered earlier. This is essentially the same as an omnichannel approach.

For sustained success, an omnichannel approach to SEO will be needed, one that accounts for both traditional and AI-driven search. But not only search, optimisation extends to other platforms and algorithms as well, including social media.

Gone will be the days when SEOs could rely on customers looking to be found on Google alone.

As things stand today, SEO is very different to what it was a year ago, let alone ten or twenty years ago. Search engines are much cleverer at spotting spam and manipulation, and they already utilise AI to filter out this kind of thing and provide more relevant results.

For most of us, this means an omnichannel approach or form of integrated digital marketing isn’t just the future – it’s the present.

SEO Will Need to Adapt

Notice my new acronym above, GAISEO (Generative AI SEO)? It’s a bit long for my liking and not as punchy as AO or even AIO (Artificial Intelligence Optimisation), still, I’m amazed nobody has coined it yet.

To ride the AI wave, both on-page and off-page SEO will have to adapt and this means taking generative AI of all types into account.

Imagine a future where AI search could understand video, image, and audio content as easily as text. Actually, we’re almost there with the latest ChatGPT, which can understand the content of images, including text and context, and provide meaningful analysis. Only last month, we got the news that ChatGPT can now see, hear, and speak.

It will only be a matter of months before this also extends to video.

SEO strategies will need to adapt to these formats as well as text.

Currently, SEO means ensuring images have sensible file names, ALT tags and captions, and including keywords where possible. For audio and video, it’s a similar process.

With generative AI, it’s now also going to be important to optimise the content of image, video and audio media, to ensure discoverability.

Grand Conclusion: Will AI Replace SEO? Only With AGI

Wow, I’ve just realised that my article looks more like a whitepaper, and I’m still writing…

To wrap up, let’s revisit the below questions:

• Can SEO be killed?
• Will AI replace SEO and ultimately search engines?
• What does the future of SEO look like?

But before doing that a few thoughts on AGI, as this may change everything I’ve said throughout this article.

If you recall, I said:

“an AGI would be able to make reasoned judgments based on the data it has, which would be continuously updated.”

The general theme throughout this article is that SEO can’t be killed but it will evolve.

AI in its current form will not replace SEO due to real-time indexing issues and a lack of real intelligence.

However, I made a prediction that AGI will happen and this will solve the real-time dilemma. An AGI ChatGPT, perhaps GPT-5, would change everything. Not only could this mean that an AGI ChatGPT replaces Google with its ability to crawl the web, index data, and update its knowledge base continuously, but it would also mean the real death of SEO.

Whether we like it or not, an AGI ChatGPT would mean Google is dead, SEO is dead, and so much more. It also seems like a possible endgame for OpenAI, meaning they would become the champions of search.

Such an outcome would most likely lead to the end of many other things, including web design and other human services that AI can automate with human-like intelligence.

Within the world of marketing, the only services that I predict will be in demand are those that require the human touch. For example face-to-face interactions, human intuition, originality and the ability to dream and think bigger than what’s done before.

So, coming back to below:

• Can SEO be killed? 
• Will AI replace SEO and ultimately search engines? 
• What does the future of SEO look like? 

The answer to all of these questions is fairly simple.

The short-term future for search engines and SEO is exciting, with lots of changes in store and new innovations. Neither will be getting killed or replaced with AI and LLMs as they exist today. The answer is very different if AGI happens.

In an AGI world, there would be no need for traditional search engines. There would be no need for SEO with AGI solutions to support businesses in their quest to get found. Web design, coding and content creation would most likely be carried out by AGIs with their ability to truly understand what’s needed and continually analyse, adapt and improve upon what’s been done before.

It’s almost impossible to imagine this world but I honestly don’t think AGI is far away, most likely starting in 2025-2026. The ramifications of AGI are likely to be wide and fast, hitting professions quickly and seeing the above scenario play out in the years leading up to 2030.

 

If you’ve made it to the end, what are your thoughts on the above? I would love to get your feedback, so please drop a comment below or connect on social media and let’s see where the conversation takes us.