Apparently SEO is dead - at least that’s what I’m seeing in approximately 90% of posts on my Linkedin feed right now. It’s been replaced by some new things - namely GEO and LLM optimization. I’ve also heard the term “Answer Engine Optimisation” bandied about.
But then Google Trends must be broken, right? It’s telling me that SEO searches are actually increasing above and beyond levels ever seen before?
The purpose of this article is to dig into these different terms, understand how they relate to each other and demystify the process of getting your brand and website mentioned in modern platforms like ChatGPT, Claude, Perplexity et al.
Common Misconceptions about the Relationship Between Generative AI & SEO
Before we get started, I would like to clear up 2 common misconceptions in relation to Generative AI & SEO:
- Chat GPT (and other Generative AI Platforms) are NOT Search Engines
- Google is NOT Losing Market Share to Generative AI Platforms
Chat GPT (and other Generative AI Platforms) are NOT Search Engines
There is fairly significant overlap between the use cases of Generative AI Platforms like Chat GPT and search engines like Google. But it’s important for businesses to note that they are not the same thing. Generative AI Platforms in their current form are “do everything machines”. Sure, some people use them for traditional searches that they might previously have used Google for, but I’m always amazed by how people overestimate this portion of ChatGPT usage. SEMrush recently carried out a study based on 80 million lines of global clickstream data to identify how people interacted with ChatGPT.
They found that only 30% of ChatGPT prompts in their dataset could be identified as fitting into one of the traditional four categories of intent associated with traditional search engine queries (navigational, informational, commercial, transactional).
The other 70% consisted of unique queries rarely seen in standard search engines. This indicates that many ChatGPT queries represent entirely new types of intent, potentially related to problem-solving, brainstorming, or exploratory inquiries that don't fit neatly into traditional categories.
This distinction is further emphasized when we look at the average length of user queries on ChatGPT - another Semrush study suggests that the average length of a user query on ChatGPT is 86 words. Compare this to the average length of a query on Google search in the USA - which is 3.4 words.
Clearly, ChatGPT and other Generative AI Platforms are being used for a much broader set of use cases than traditional search - from creative writing, to coding, to summarising documents, to organising tasks, and so on.
Google is NOT Losing Market Share to Generative AI Platforms
With all of the hype around Generative AI Platforms (ChatGPT in particular) and the latest “death of SEO”, you would be forgiven for assuming that these platforms are aggressively eating into Google’s market share.
However, you would be wrong.
According to a SparkToro study in March 2025, Google saw more than 5 trillion searches in 2024, or about 14 billion per day, giving it a 93.57% market share.
ChatGPT saw an estimated 37.5 million search-like prompts per day, giving it a 0.25% market share. That’s less than Microsoft Bing (4.10%), Yahoo (1.35%), and DuckDuckGo (0.73%).
Even more recent data from SEMrush suggests that the framing of ChatGPT vs Google is a false dichotomy, pointing out that when new users adopt ChatGPT, their quantity of Google searches spikes, and remains higher-than-before even months later.
All that being said, usage of ChatGPT and some other Generative AI platforms is likely to continue to rise as long as the astronomical investment in these products continues. Usage and market share estimates vary wildly depending on the source. The reality is that no-one can predict the future - we may see Generative AI platforms completely taking over from search engines and I may end up eating my words, or the Generative AI bubble may burst and we may end up back at square one with a more AI-focused version of the same old Google monopoly.
Regardless, whilst it’s not wise to redirect all your focus into “ranking” within Generative AI platforms, it’s equally unwise to ignore them entirely. They are a new channel that is very much worth your consideration, and as we will outline throughout this post - optimising for Generative AI platforms is likely to have a positive impact on your search engine rankings and vice versa. By tweaking the focus of your SEO strategy and increasing the focus on a couple of tactics, you should be pretty well covered in terms of GEO.
Time for Some Definitions
So now that those clarifications are out of the way, let’s move on to some some definitions:
SEO (Search Engine Optimisation): The practice of improving a website’s visibility and ranking in traditional search engines (Google, Bing, etc.) through technical optimisation, content creation, and authority building.
GEO (Generative Engine Optimisation): The practice of optimising for Generative AI Platforms (ChatGPT, Gemini, Claude, Perplexity, etc.) so your brand is cited, recommended, or linked in AI-generated responses.
LLM Optimisation is another term that I will define after providing some more context.. It’s a term often used interchangeably with GEO but technically they are not the same thing because technically an LLM and a Generative AI Platform are not the same thing.
What’s The Difference between an LLM and a Generative AI Platform?
People often use the terms interchangeably but there is a distinction to be made between “LLM” and “Generative AI Platform”.
An LLM (Large Language Model) is an AI modeI focused specifically on understanding, processing, and generating text. LLMs are essentially massive statistical engines trained on huge amounts of text (books, websites, articles, code). They predict the most likely sequence of words, based on patterns it learned during training.
A Generative AI Platform is a system built on top of LLMs (and other generative models), providing the tools, infrastructure, and integrations businesses and users need.
ChatGPT, Gemini, Claude, and Perplexity are all generative AI platforms built on large language models. They go beyond the base LLM by adding features such as “grounding” via web search (in some cases), multimodal inputs (text, images, audio, etc.), and user-friendly interfaces.
With fine-tuning and external grounding, these platforms become tools that feel like they can “understand” and “reason,” even though it’s fundamentally still just pattern prediction.
What about Google’s AI mode and AI Overviews?
Google has also introduced its own generative AI experiences inside Search, most notably AI Overviews (AIO) and AI Mode. These are not standalone Generative AI Models but rather Generative AI features within existing products.
AI Overviews (AIO): A feature within Google Search that generates concise, AI-written summaries at the top of results pages, often with links to supporting sources. These summaries are powered by Google’s Gemini LLM, grounded in web content retrieved from the index.
AI Mode: A newer interface within Search that blends traditional results with generative answers. When users switch to AI Mode, they are effectively interacting with Search powered by Gemini, rather than just the classic ranking algorithm.
While the technology under the hood is similar to generative AI platforms like ChatGPT or Perplexity, there’s one crucial difference: to be included in AI Overviews or AI Mode, your content first needs to be visible in Google Search.
According to Google themselves:
“You can apply the same foundational SEO best practices for AI features as you do for Google Search overall: making sure the page meets the technical requirements for Google Search, following Search policies, and focusing on the key best practices, such as creating helpful, reliable, people-first content.”
Both AI Overviews and AI Mode function slightly differently to traditional search results though in terms of information retrieval. They both use an information retrieval process called “query fan-out”.
What is Query Fan-Out?
Query fan-out (sometimes called multi-query expansion or search fan-out) is when a Generative AI system takes your single prompt, rewrites it into multiple sub-queries, sends those out to external sources (like web search or databases), and then combines the results before generating an answer. Olaf Kopp explains the purpose of query-fan out very well:
“The core function of query fan-out is to transform a single, potentially complex or ambiguous, user query into a multitude of more specific sub-queries.”
According to Google, AI Mode and AI Overviews…
“may use a "query fan-out" technique — issuing multiple related searches across subtopics and data sources — to develop a response. While responses are being generated, our advanced models identify more supporting web pages, allowing us to display a wider and more diverse set of helpful links associated with the response than with a classic web search, enabling new opportunities for exploration.”
You can read more about query fan-out in Google’s AI mode in this excellent article by Marie Haynes.
The truth is that some form of query fan out has been used by Google for years even for “pre-AI” search. Other search-centric platforms like Perplexity and Bing Co-pilot also rely heavily on query fan-out. Chat-centric platforms like ChatGPT and Claude use it for queries that can’t easily be answered via their training data - in other words the queries that require “grounding”.
What is “Grounding”?
LLMs are powerful but have two big limitations:
- Knowledge cut-off: they don’t know events or data after a certain date.
- Hallucinations: they sometimes generate fluent but factually wrong answers.
Grounding is an attempt to counteract these limitations by pulling in relevant, authoritative information at the time of the query. Grounding essentially means connecting an LLM’s answer to real, external data sources (usually via web search) instead of relying only on what the model learned during training.
A platform like ChatGPT or Claude typically won’t use grounding for straightforward questions with “permanent” answers like “Where was Erling Haaland born?”.
However, for a search like “Who is currently the top scorer in the 2025/26 premier league?”, the platform will use “grounding” in the form of query-fan out across the live web to ensure that they can provide an accurate and up to date answer to the query.
Let’s Revisit Our Definitions
So now that we understand this context and the relationship between LLMs and Generative AI Models, we have a clearer distinction between GEO and LLM optimisation.
LLM Optimisation: The practice of ensuring your brand, expertise, and content are well-represented in the training data of Large Language Models (e.g. ChatGPT, Claude, Gemini), so the model “remembers” and generates responses that reference or recommend you even without live grounding. This relies on building widespread, high-authority, high-visibility signals across the open web that are likely to be captured during model training.
GEO (Generative Engine Optimisation): The practice of optimising for Generative AI Platforms (e.g. ChatGPT, Claude, Gemini) so your brand is cited, recommended, or linked in AI-generated responses. This encapsulates LLM optimisation but also includes influencing grounded responses that draw from live information across the web. GEO therefore covers both the long-term objective of being embedded in model training data and the short-term objective of being surfaced when platforms ground their answers in real-time web content.
Read that second objective again… We want to optimise our website and brand to be:
“surfaced when platforms ground their answers in real-time web content”.
Sounds familiar, right?
Replace “platforms ground their answers” with “users search for answers”
This, in effect, shares the exact same objective as traditional SEO - we want our brand and our content to be surfaced prominently when users or bots search the web for relevant queries.
So what does GEO involve?
To be honest, I’m hesitant to even use the term GEO due to how it tends to be used by a certain group of people - i.e. fearmongers, charlatans and frauds who position it as a brand new discipline, completely distinct from SEO in order to upsell “new” services.
I tend to share the frustrations expressed by Catherine Lux, Head of SEO at Assembly global below (yet here I am writing a 4,000 word article on GEO…):
The reality is that at a tactical level GEO is a subset of SEO and, in fact, the tactics required to optimise for Generative AI Platforms are much the same as those required in traditional SEO with an increased weight on a few specific tactics. If you’ve got a good SEO agency, consultant or in-house team who has been doing all the right things over the past number of years, then you can rest assured that this has been having a positive impact on your visibility in Generative AI Platforms and Google's AI Features as well.
If we must use the term, then the GEO equation, as I see it, is as follows:
GEO = LLM Optimisation (the objective is to get into the model’s training data so it’s “baked in” and recalled even without grounding) + Grounding Optimisation (the objective is to be visible to Generative AI bots for relevant “grounding” queries on the live web - essentially this is traditional SEO with some slight tweaks).
There is a tonne of overlap between both of these “pillars” when it comes to the actual tactics required to achieve both objectives. As is the case with Traditional SEO, we can still pretty neatly segment these tactics into 3 categories - Technical Optimisations, Content Optimisations and Authority Optimisations (again with some overlap).
Technical Optimisations for GEO
Ensure that Your Website is Accessible to “AI Bots”
From a technical perspective, the first thing you want to do is ensure that your website and content is actually accessible to bots. Without access to your site’s content, generative AI platforms (ChatGPT, Gemini, Perplexity, etc.) cannot surface your brand in responses.
Many of you may feel conflicted about this and for good reason - the ethical and copyright concerns associated with allowing these companies to train their LLMs on all of the information on the open web are significant.
On the flip side, visibility in Generative AI platforms depends on those systems being able to access, crawl, and learn from your content. Blocking their crawlers risks your brand being invisible in AI-generated answers (at least those that don’t require grounding), which could reduce discovery and traffic in the long term.
You’ll have to weigh up this conflict and decide what the best approach is for your business and your website(s) - for some businesses it will be less of a concern than others. Typically, these platforms will have different bots for search and agent use vs AI training data collection so you could make the call to keep your content out of the training models but allow crawlers to access your site when grounding - e.g. the example below from Search Engine Land:
Another accessibility consideration is content formats - don’t hide key assets in PDFs or gated formats only. HTML pages are far more likely to be ingested.
Use Clean, Structured HTML
To maximise visibility to these bots, you should use plain HTML for core content and maintain a logical HTML structure with basic SEO tags (<title>, <meta description>) and semantic tags (<h1>, <h2>, <p>, <ul>, <article>, <section>, <nav>), making it easier for LLMs to parse hierarchy and meaning.
You should also avoid hiding key content behind interactions like accordions, tabs, or “read more” buttons that depend on JS. Many generative AI bots don’t handle JavaScript rendering well, if at all. Content locked behind dynamic loading, infinite scroll, or scripts may never be ingested. Most AI bots remain relatively lightweight and only parse the raw HTML.
And don’t forget about links - these shouldn’t rely on JavaScript either. Most of these platforms rely on search engine APIs to crawl the web during grounding - and most search engines find it much easier to crawl html links rather than those that rely on Javascript.
Use Schema.org Structured Data Markup (Controversial Inclusion)
So… let’s talk about schema. Whether schema.org markup has any direct impact on LLM visibility is a topic of vociferous debate in SEO circles. Given how LLMs work (essentially predicting words and patterns rather than parsing structured data in the same way a search engine does), it’s entirely possible that they just interpret schema as text like any other part of the page.
That said, I still think it’s worth implementing for two key reasons:
- Search engines use it: Schema absolutely influences how Google (and others) understand your site, which directly affects whether your content gets pulled into AI Overviews and other grounded features. If you want to maximise your grounding visibility, schema is part of the toolkit.
- It’s written for bots: Schema.org exists for one explicit purpose: to make it easier for machines to understand your content. Even if LLMs aren’t fully utilising it now, there’s no harm in future-proofing. If and when models start leaning on structured data more directly, you’ll already be covered.
So, while you’ll hear heated arguments about whether schema has any impact on LLM optimisation, in my opinion it can’t hurt - it helps in SEO already, and it may prove valuable down the line. That’s enough of a business case for me.
My recommendation would be to add detailed organization schema markup to your home page, add article and author markup to your blog posts, add relevant product data to your product pages and product collection schema on your product listing pages.
Content Optimisations for GEO
Create Comprehensive Content Guides and Hubs
Good SEOs have been creating topic clusters and content hubs for years. Covering a subject comprehensively makes it easier for both search engines and generative AI systems to understand your authority and pull your content into answers.
Side note: For as long as I remember I’ve encouraged clients to prioritise comprehensive “guide” style content, broken down into sections (based on topics) and sub-sections (based on sub-topics) with a logical flow. If you want to rank #1 on Google for questions related to a certain topic, then you should ensure that you have the most comprehensive, well written, well cited (more on that later) and well structured content on that topic on the entire web. If it’s not the best, then why should we expect it to rank?
GEO charlatans will rebrand this as “query fan-out optimisation” but really, it’s just a solid content strategy. If your site already has well-structured pillar pages supported by in-depth cluster content, you’re in a strong position.
If you’re not doing so already, make sure that you’re doing comprehensive topic and keyword research for every guide or blog post that you write - cast the net wide and make sure you’re covering as many potential user queries about the topic as possible. After you get these into an outline you’ll have a decent idea whether it requires one comprehensive guide or a pillar and series of clustered pages. Tools like Also Asked and SEMrush’s Keyword Magic tool are useful for this, but good old manual SERP research can’t be underestimated.
Use Logical Content Formatting and Structure
I’ve touched on formatting and structure above but it warrants its own explanation. Generative AI systems need to extract answers cleanly. Well structured content and logical formatting like clear headings and subheadings, short paragraphs, bullet points and numbered lists improves the likelihood of your content being surfaced.
Each individual topic within your guide / post should live under a H2 and should be broken down into bite-sized smaller sections (under H3s) where relevant. LLMs often look for concise, quotable statements that can be lifted into answers. Including FAQ sections, featured-snippet-style definitions, and clear explanations of key terms will increase your chances of being cited. This isn’t new - SEOs have been writing for featured snippets for years but the payoff now extends into generative AI visibility.
Our friends the GEO charlatans have started calling this “content chunking” or even “optimising for vector processing.” Feel free to ignore the buzzwords. This is simply good writing and web publishing practice: break down complex topics into scannable, structured sections that machines (and humans) can digest easily.
Keep Content Fresh and Relevant
Many experts argue that grounding systems tend to favour fresher sources, with ranking systems in RAG architectures often giving higher weight to content that is more recently published or recently updated. While this isn’t a universally established guarantee, it aligns with best practices: if your content is stale, it’s less likely to be pulled into grounded responses.
Don’t just publish once and forget - keep content living and breathing. You should also include visible dates and <meta> tags to help AI understand when content was published or updated (okay this is a technical recommendation again but it fits better here).
Authority Optimisations for GEO
Off-Site / Brand Building
Inclusion in training data is heavily influenced by brand visibility and footprint across the wider web. Particularly when it comes to searches with a transactional / commercial intent (e.g. “What is the best software for X/Y/Z?”) LLMs like to hedge their bets by basing their answers on the consensus within their original training data.
If you’ve been working on an SEO strategy up to this point, you’ll probably have been focused on some of this anyways, but in the age of LLMs and Generative AI Platforms, this is arguably more important and impactful than ever. To boost your brand visibility consider tactics like:
- Digital PR & outreach
- High-authority links & citations
- Guest contributions
- Directories & review platforms
- Academic & government references
While the primary focus of these tactics in the past was to build backlinks, the focus in the context of GEO is more concerned with relevant, contextual citations.
Platform Diversification
LLMs crawl more than just company websites and, as we’ve already mentioned, “feel” much more comfortable surfacing answers and recommendations that align with the consensus on trusted websites and platforms across the web.
You should diversify the platforms on which you publish content - contribute to platforms like LinkedIn, Medium and Substack. Many GEO experts have also been recommending a Reddit strategy - however my personal opinion is that this is a place where customers should be talking (hopefully positively) about your brand, rather than a place for brands to be self-promoting.
If you can encourage customers to post positive content about your brand on 3rd party platforms (e.g. review sites), this is an added bonus.
Same Fundamentals, But New Challenges
So technically, the fundamental optimisations required to actually appear in Generative AI Platforms for relevant queries remain very similar to traditional SEO tactics (if you were already doing SEO correctly) - arguably with an added focus on off-site / brand building.
However, whilst the tactics remain pretty much the same, these platforms and products have changed the search landscape in one major way. When it comes to informational intent queries - users are less likely to visit your website from a Generative AI Platform like ChatGPT or one of Google’s LLM driven AI features like AI Overviews. Essentially, the growth of these platforms and features is leading to an increase in “zero click searches” i.e. searches for which users never click through to a website.
When it comes to Google’s own AI features - According to BrightEdge research, in the year after the introduction of AI Overviews, the number of impressions on Google increased 49% but the click-through rate fell 30%. In another aHrefs analysis across 300,000 searches, top organic results lost 34.5% of their clicks when an AI answer was present.
I have run multiple reviews of websites that have seen a drop in CTR since the introduction of AI Overviews, and have observed that AI Overviews are mainly reducing click-through rates for a certain subset of informational-intent queries that I call "immediate response" intent content. In other words, situations where users ask Google a specific, typically low-stakes, question for which they just want a quick answer. The hallmark of “immediate response” intent queries is that the users don't really care that much about the source of the content and they have no intention of reading a comprehensive guide etc.
As for standalone Generative AI platforms, whilst things were looking quite positive in terms of click through to websites, we’ve seen a significant reversal in this trend recently with one study showing that ChatGPT referral traffic dropped by 52% in August 2025 vs July. This teaches us a valuable lesson - these platforms are constantly evolving and we cannot take for granted that any trends will persist in the long term.
In fact SEO expert, Malte Landwehr, recently shared data that suggests that zero click searches are even more prevalent on ChatGPT and other AI platforms / features than on Google search.
So whilst the tactics are familiar, the age of AI answers does require a fundamental shift when it comes to SEO at a strategic level…
The Impact of Generative AI on SEO Strategy
So the arrival of Generative AI features doesn’t mean we need to rip up the SEO playbook when it comes to optimisation tactics, but it does force us to look at strategy differently in a couple of key areas.
1. An Evolution in the Role of Informational Content
In all the AI Overview impact reviews that I’ve done, the pages hit hardest by “immediate answer” queries (the ones where users just want a quick, low-stakes fact) are still among the top organic traffic drivers for those sites.
Why? Because plenty of people still want comprehensive, well-cited, expert-led guides. When someone is genuinely researching a topic, they’ll still click through to long-form content. But if they’re just after a definition or a stat, you’ll need to accept that the click is less likely. On the upside, if you’re cited in an AI Overview, your brand still gets visibility even without the click.
Interestingly, one client I reviewed actually saw an uplift in Bottom of Funnel (BOFU) content since AIO rolled out. When I considered the SEO strategy that we had been following, this made sense - we had been building topical blog clusters of TOFU informational content around those BOFU pages, strengthening their authority. So even if informational content delivers fewer clicks than before, it still plays a critical role in boosting rankings and visibility for related service, product, or industry pages.
What does this mean for your SEO strategy?
You should keep creating expert-driven, well-structured, well-cited long-form content. Yes, you’ll probably get fewer visits from the “immediate answer” crowd, but that traffic was rarely high-converting anyway. Informational content still plays a key role in:
- Authority building: expert-driven informational content that demonstrates EEAT still plays a major role in boosting the overall visibility of your site.
- Brand visibility: being referenced in AI answers (brand logo and name front and centre) gets your brand name in front of relevant eyeballs regardless of whether or not the immediately click through
- Positioning - expert-driven informational content that answers your target audiences queries or solves their problems positions you as a trusted source in your space.
- Traffic - whilst the traffic levels for your informational content is likely to be lower than it would have been previously, it will still generate traffic itself as well as boosting visibility and traffic of the product/service pages around which it is clustered
At the end of the day, if you want your service or product pages to rank, it’s far better if they sit within a wider cluster of guides and resources that showcase expertise than to leave them standing alone without context.
2. Bottom-of-Funnel (BOFU) Content is More Important Than Ever (or at least as important as it always should have been)
Bottom of Funnel Content should always have been high on your list of priorities. But in reality, many companies got distracted by clicks and sessions metrics associated with broad informational content despite much of this traffic never having any intention of converting - they were just looking for an immediate answer.
Now more than ever, you need to ensure that you have comprehensive, well optimised content on your website for bottom of funnel searches. That means:
- Product / Service / Feature pages: you should have a specific page for each product / service / feature, NOT just one page that lists them all. By having a single, comprehensive, well optimised page for each product / service / feature you are increasing the likelihood of being surfaced in Generative AI platforms and Google for searches related to that specific product / services / feature.
- Industry-focused pages: you should have individual pages dedicated to each of your target industries / audiences. These pages should outline the benefits of your product or service to the specific target audience and include examples, testimonials and other trust marks. This will increase the likelihood of your website or brand surfacing in answers to queries about the most relevant products or services for those audiences.
- Product Listing Pages (PLPs): For eCommerce sites, PLPs or Product Category Pages are an oft-overlooked opportunity when it comes to SEO and GEO. Ensure that you have a well thought out site hierarchy with PLPs dedicated to all categories of products for which there may be high search demand amongst your target audience.
So informational content is still worth investing in, but more for the sake of authority and brand positioning than as a direct traffic play. The real strategic shift is making sure your BOFU content is airtight - because that’s where the clicks and conversions are going to come from in the age of AI answers.
A Tangentially Related Rant
I hope you’ve found this guide helpful so far. From this point onwards, you won’t learn anything else about SEO or GEO, but I don’t feel comfortable writing such a detailed post about these technologies without also discussing some of the concerns that I have in relation to them.
Platforms like ChatGPT, Perplexity and Claude are useful tools for marketers. They can help us to summarise content, write blog posts and generate ideas for campaigns, social posts, or client pitches.
Their usefulness and importance however is, in my opinion, being blown out of all proportion by a hype-cycle the likes of which has never been seen before. The idea that these platforms are a route to AGI that is going to “solve physics”, “cure cancer” or lead to “free global education” (all of which have been claimed by Sam Altman at different stages) is at best unrealistic and at worst misleading and if the current unprecedented levels of money and power that is being funnelled into these companies continues to flow, I fear that things will not end well.
It seems to me that an over-reliance on these tools is not only a bad idea from a business perspective by limiting creativity and the likelihood of novel ideas, it is also bad for our brains, the economy and the environment.
Cognitive Impact of Over-Reliance Generative AI Platforms
At this stage there are countless studies (like this, this , this and this) that outline the link between the use of generative AI platforms and the erosion of cognitive abilities like critical thinking, problem-solving skills, and memory retention.
If you’re attuned to the writing style of ChatGPT and other Generative AI tools, a quick look through your LinkedIn feed or even a sift through your email inbox with an inquisitive eye is likely to turn up plenty of examples of comments or replies that are suspiciously “robotic”.
Worse than this, I can think of numerous examples of real-life discussions in which I’ve questioned someone’s opinion or a “fact” or “figure” that they have brought up and they have admitted that they “asked ChatGPT”. Did they corroborate any of the information provided to them by their LLM buddy? Usually not.
People are outsourcing their thinking to tools that cannot think. If you were to explain this to someone in the 1950’s or even the 90’s or 00’s it would seem absolutely absurd. So why are we so unmoved by it now? In my opinion, this is an incredibly slippery slope.
Economic Impact of Generative AI Platforms
The level of investment into the Generative AI industry is unprecedented. Consider these facts and figures:
- According to some reports, nearly half of the world’s private investment is being funnelled into AI, and AI speculation is the main driving force behind the S&P 500’s recent growth.
- In terms of VC investment, according to a recent report by law firm Ropes & Gray, AI-related investments accounted for the majority of VC deal value in the first half of 2025 (51%), compared with just 12% of total VC deal value in 2017.
- A recent Generative AI Key Deals and Market Insights study by EY reported that global venture capital (VC) investment in generative artificial intelligence (GenAI) surged to $49.2 billion in the first half of 2025, outpacing the total for all of 2024 ($44.2 billion) and more than double the total for 2023 ($21.3b), according to.
- Earlier this year, OpenAI closed a $40 billion funding round - the largest private tech deal in history. On September 22, Nvidia confirmed plans to invest a further $100 billion in OpenAI as part of a deal that includes plans to build and deploy upwards of 10 gigawatts of AI data centers (more on that later).
- Mark Zuckerberg has said that Facebook will invest at least $600 billion through 2028 in U.S. infrastructure to support its future plans to which Generative AI is central. A number of days later, around the same time that Meta introduced Vibes - their new AI slop generating machine - Zuck claimed that “if we end up misspending a couple of hundred billion dollars, I think that that is going to be very unfortunate obviously”. Unfortunate indeed.
There are plenty of experts and commentators online that have been calling this a bubble for. Now mainstream tech and finance media appear to be following suit - with a recent Fortune article claiming that the AI boom is unsustainable unless tech spending goes ‘parabolic’.
I can’t help but think that there are better uses for all this capital than these general purpose, energy intensive, do everything kind of well but do nothing properly machines. And if / when the bubble does burst, many experts believe it would trigger the economy into a recession.
According to the economist, Noah Smith “it could even lead to a financial crisis if the unregulated “private credit” loans funding much of the industry’s expansion all go bust at once.”
Environmental Impact of Generative AI Platforms
Remember those plans for 10 gigawatts of AI data centers? Well, according to Cornell University energy-systems engineering professor Fengqi You: “Ten gigawatts is more than the peak power demand in Switzerland or Portugal”.
According to Brid Smith on the excellent Tech Won’t Save Us podcast back in December 2024 - data centers were at that point responsible for “23 or 24%” of energy consumption here in my home country of Ireland. I can only imagine that figure is growing.
According to an MIT article, Scientists estimate that by 2026, global data centre consumption is forecast to hit 1,050 terawatt-hours. If data centers were a country, that would make them fifth largest electricity consumer in the world, ahead of Russia and just behind Japan.
The ever expanding requirements for more and more data centers and the vast amounts of energy required by these companies if they continue on the current trajectory is a cause for major environmental concerns in relation to electricity demand, carbon emissions and water consumption in particular.
Akepa have put together the infographic below based on their own research. I’m no environmental expert, but it makes for grim reading.
Have We Been Sold a Pup?
Generative AI is just one type of AI. There are many other types of AI - I’m not an expert on any of them but I don’t believe that it’s beyond the realms of possibility that they might actually contribute to solving some of the very real issues that Sam Altman and those pushing the Generative AI hype-cycle deceptively claim that Generative AI will solve.
Generative AI tools though are where the money is going, not because anyone necessarily believes these claims, but because these are the tools that are promising “efficiency” to business owners. The end goal of course is to become so efficient that they can “cut costs” (fire humans).
Those efficiencies though, in reality, don’t seem to be coming to fruition - recent MIT research finds that whilst companies are investing billions in AI tools like ChatGPT, aiming to reduce costs and boost productivity, 95% of businesses using AI aren't making their money back.
Final Thoughts
I’m aware that I’ve covered a lot in this post and gone off on a tangent towards the end. And it feels a bit jarring to close this post by recapping my “top tips” for Generative AI Platform visibility - but that’s reflective of the awkward position a lot of us SEOs are in. It’s our job to help brands to drive organic online traffic, which of course includes traffic via Generative AI platforms. But the closer we study these platforms and their impacts on the world, the more we realise the risks that they pose.
That being said… Here are my top tips for Generative AI Platform visibility 🥲:
- Continue doing good traditional SEO
- Double down on BOFU content
- Build brand presence and authority across the web
These are the tactics that are most likely to stand the test of time, no matter how the platforms evolve.
And stop using ChatGPT as your brain!
I'm Darren, founder of SEO by Darren. I develop bespoke SEO roadmaps and implement long term SEO strategies to build organic visibility, traffic and conversions for my clients.
