When we want to learn about a certain topic, we often start from Googling it or asking AI to tell us what it’s about. Search engines and LLMs will provide a result that will look very authoritative, but there is a certain logic to what’s going to be generated and shown at the top. Knowing the rules of this logic can help us regain control of the information that we consume and help us introduce our own ideas into any discourse.

Both search engines and AI chatbots strive to provide the best possible answer. They do so by forming a general understanding of all possible topics that a good answer would include. Then they will find the search results or generate a response that would cover as much of those topics as possible. Usually, it’s quite effective, but there can be a problem if their understanding is incomplete or biased towards a certain topic. At the same time, it also makes them vulnerable to infiltration: if we create the content that touches upon all the important topics and concepts and — at the same time — links them to a notion or agenda we want to promote, we can introduce our ideas into the public discourse in a very effective way.

Search engine optimization and marketing specialists know how it works very well. There are multiple tools that help them do that at scale. Therefore, it is important to know how it works not only to avoid being manipulated but also to know how to promote your ideas and ideals. These techniques should not be limited only to marketing and politics. Every individual should have the capacity to introduce their ideas into the public discourse and be heard. In this article, we’re going to show it works.

1. Defining the Objective: from Singularity to Polysingularity

Let us start with a real example. There is a well-known concept of singularity: the point where technological advancement and AI reach the point of no return, leading to uncontrollable growth and profound changes to civilisation. Some people fear it, some can’t wait for it to occur. There’s something religious, apocalyptic, and prophetic about it — all at the same time.

For me personally, it’s always seemed like singularity lacked some multiplicity in it. It feels too totalitarian, too specific, too much like a single point rather than a multidimensional shape. Therefore, I coined up the term “polysingularity” to describe an alternative version of reality: striving towards the multiple truths, diversification of ideas, preventing totalitarianism of thought and ensuring that there is a certain degree of separation and collaboration at the same time — qualities that usually are displayed by resilient ecological systems.

I want to promote this idea to those who think about singularity. How do I do that?

2. Hacking the Search Engine

The first step is to understand the existing discourse on the topic of singularity. Or, rather, what people find when they search for it. According to the recent study conducted in 2024 by The National Bureau for Economic Research, more than 80% of the people in the USA and EU use Google as the first point of entry when they look for something.

Let’s visualize the results that we get when search for “singularity” on Google using the InfraNodus visual search app. InfraNodus extracts the top 10 pages in the search results and then analyzes the content of the pages and visualizes them as a knowledge graph. The concepts are the nodes, co-occurrences are the connections. We will then see the main topical clusters and ideas related to the concept of singularity.

If we create the content that spans all those topics, then Google will consider our contribution to be of a high quality and will push it to the top of the search results. While there are other factors at play (domain rating, backlinks, article length, informational gain), which are well-known to SEO marketers, my experience shows that content still matters and I’ve pushed many (often esoteric) ideas to the top of the search results using this approach (check “polysingularity” or “bodymind operating system” for instance).

The main topics related to “singularity” are:

1. Intelligence surge (a dramatic improvement of human and artificial intelligence)
2. Tech evolution (technology evolves and becomes much better)
3. AI scenarios (how AI will evolve)
4. Machine mind (merging with the machine)

The main concepts related to “singularity” are:

human, intelligence, AI, technological

This is how Google sees the concept of “singularity”. This is also the terms that LLMs and AI chatbots will associate the concept of “singularity” with because they train on the web data.

And this is how you will see it too, because you’re most likely to use Google or ChatGPT to find out more about it.

Moreover, according to the recent study of click-through rates published by a popular SEO blog Backlinko, more than half of the clicks will go to the first 3 results. Let’s filter the graph to highlight the part that the top 3 pages cover:

As we can see, the first 3 results cover almost all the graph, all the topics and concepts in the discourse.

Search results at the positions 4 to 7 are quite niche: they touch upon peripheral concepts:

Finally, the results at the positions 8 to 10 cover the whole graph again:

This yields some very interesting insights:

1. Google will favor the pages that cover most of the content, topics, and concepts covered in the first 10 websites that it returns for a search query (positions 1 to 3)
2. However, it will also introduce the pages that talk about more niche and specific aspects of the topic to the middle of search results (positions 4 to 7)
3. Finally, it will show the rest of the content and it will give preference to the pages that cover all the topics.

So what makes it choose to show some pages at the top of its first page (positions 1 to 3 that will get 50% of the clicks) in favor of the 3 last pages at the bottom of the page (that will only probably get about 5% of the clicks)?

Apart from the well-known criteria such as the quality of the website and the number of other high-quality pages linking to it (usually referred to as domain rank) there are also other additional factors at play.

We can add additional filters in InfraNodus and ask it to show us the <h2> titles (used to indicate headlines for the sections inside the text) for the pages at the first 3 positions:

As we can see, the pages at the top position use the terms “explosion of intelligence”, “mind program” and “innovation” in their titles.

For the pages at the last positions we see a completely different picture:

They are leaning towards technology and future.

Therefore, Google prefers the pages that have more general section titles and exhibits a certain bias towards the link of “singularity” to the “explosion of intelligence” and “mind program” rather than the technological aspects.

Therefore, if we create the content that covers the topics we outlined above, but make a particular emphasis on the notions of “explosion of intelligence”, “superintelligence” and what it does to our minds (in the text section headlines using <h2> tags), we will create a high-quality article on the topic that will be pushed to the top of the search results.

If we then want to introduce the idea of “polysingularity” into the discourse on “singularity”, we can do that by linking it to those concepts and injecting it into various parts of the text. Another strategy is to establish authority on the topic of “singularity” first and to then create adjacent content linking to “polysingularity”.

3. What’s in an LLM? Peeking into the AI’s Brain.

To verify our findings, we can also analyze what LLMs think about the concept of singularity. We can do that using the InfraNodus AI content generator. This app will build a knowledge graph based on the responses auto-generated by GPT-4o using the same algorithm as the one outlined above.

For “singularity” we get the following representation of ideas related to it. This graph actually represents the most likely concepts for LLMs to generate when they are prompted with the concept of “singularity”, so in a way get to peek into a simplified (2D) representation of the multidimensional AI brain:

As we can see, the main concepts are quite similar to the top Google search results:

human, singularity, technological, intelligence

The main topics are:

1. Technological singularity (similar to topic 2 in Google)
2. Human intelligence (same as topic 1 in Google)
3. Philosophical inquiries (here GPTo is a bit more dreamy than Google)
4. Exponential change (similar to the topics in Google)

We can now overlap the graph of AI-generated results and the Google search results and verify that they are quite similar to one another:

The “Content Gaps” panel in InfraNodus shows us that what AI is missing (comparing to the Google search results) is the discussion of how it would be integrated with the human brain and the various scenarios related to that.

Maybe it’s hiding something? In any case, that would be an interesting point to address. So when we talk about “polysingularity” in relation to these topics, it could be interesting to mention its effects on the brain as well.

4. Informational Gain: Bridging the Content Gaps

Positive contribution to a discourse will often imply that we:

1. cover the topics that are important for this discourse (what we discussed in the previous sections)
2. bridging the gaps between them in a new way or going beyond the periphery

In fact, Google has a patent on so-called “informational gain”, where they talk about showing the user the content that adds the most knowledge in relation to the documents they’ve seen before. Practically this means that the content that links the already known topics in a new way will be favored by Google and shown at the top of the search results, because it will make a unique contribution while touching upon the ideas considered to be important for the discourse.

InfraNodus has a structural gap detection algorithm that analyzes the text knowledge graph generated from Google search results and shows the gaps that exist inside:

We can generate a few of them using the InfraNodus 3D graph browser extension:

Gap 1:
“38%: Digital Intelligence “,”human; intelligence; artificial; explosion; create; understand; future; result; brain; agi; humanity; superhuman; development; level;”
“7%: Machine Design “,”machine; mind; science; computer; suggest; design; robot; achieve; intelligent; process; economic; discuss; social; replace; capable;”

Gap 2:
“38%: Tech Singularity “,”singularity; technological; technology; time; accelerating; progress; growth; advancement; innovation; kurzweil; argue; rate; version; involve; change; people; leading; continue; article; scientific; reach; world; life; vinge; complexity; current; evolution; model; energy; point; concept; context; believe; exponential; post; approach; acceleration; history;”
“7%: Machine Design “,”machine; mind; science; computer; suggest; design; robot; achieve; intelligent; process; economic; discuss; social; replace; capable;”

These gaps show that there’s a lack of connections between machine design (and AI hardware) to the topics of singularity and intelligence. Perhaps, we should ask ourselves how would this be technically implemented or built?

If we create the content that bridges this gap with the new idea (e.g. Polysingularity), then we will generate a higher degree of informational gain making this content more attractive both for Google and for AI, thus increasing the chances that the concept of Polysingularity will come up especially when our potential readers will ask how the discourse on singularity could be developed further.

5. Weaving It All Together (with Google Gemini AI)

Let us now take all those insights and weave them together into an article that we will then post online. After some time, it will get indexed by search engines and AIs and will hopefully introduce the idea of polysingularity to those who are interested in singularity.

The best way to approach is to to write the article ourselves, but for the sake of the demonstration we will show how to use Gemini Pro 2.5 to achieve the same goals. We will open the LLM interface and add the following prompt:

You need to create a SEO-friendly article that will cover the topics identified in the text below related to the concept of singularity and polysingularity.
You need to read the text below and generate a prompt that would produce the result required. Take particular notice of all the observation inside the text.
First you generate a prompt that will be needed to generate the SEO friendly article that will relate to all the topics identified below.

Then we copy the text we wrote above that has all the instructions after this prompt.

Gemini will generate a long-form prompt which we can then feed back to the model in order to generate the article that we want to create.

Once the article is generated, we can visualize it using InfraNodus text analysis app:

We can see that it covers the main topics we talked about: Future Convergence, AI, Technological evolution — that is, everything that relates to the top Google search results. But it also brings in the whole discourse on Intelligence diversity, which is, in fact, central in this discourse.

We can also visualize how it fits into the Google discourse by choosing the Compare To mode in InfraNodus and then selecting the Google results graph. The red nodes show how well the new text is integrated into the Google search results discourse:

As we can see, we cover all the important concepts and relations. However, some are also missing:

problem, increase, million, life, time, global factor

We could feed this data to the Gemini AI prompt and ask to modify the article to include these concepts as well.

After a few iterations we can publish this article on the Polysingularity blog and then once the search engines start adding it to the search results use the feedback we get to push it up further in search results based on the long tail keywords it’s going to rank for.

…

To try this approach: https://infranodus.com