Aydın Tiryaki (2026)
Introduction: From the Age of Innocence to Manipulation In the late 90s and early 2000s, when the internet was still in its infancy, building a website was akin to organizing books in a digital library. We used to meticulously place “keywords” that best described the content into the HTML codes of the pages we prepared. This was a classification method based on the “honor system” envisioned by the founders of the internet. However, before long, clever tricksters exploited this system by hiding popular but completely irrelevant words within the code, shattering this trust-based structure. This breaking point was the beginning of the process that forced search engines to evolve from simple “word matchers” into “meaning seekers.”
Forced Evolution: Search Engines Learning to “Read” To find the truth in a data ocean polluted by manipulation, Google and other search engines had only one solution: To focus on the content itself, not the human declaration (meta tags). Algorithms, which initially only looked at keyword density, eventually had to decipher the relationships between words, synonyms, and the context within a sentence. Because as spammers evolved, the engineers trying to catch them had to evolve as well. As the famous saying goes, “Necessity is the mother of invention.” This necessity—born from the information pollution on the internet—became the greatest motivation for engineers to solve the mathematics of language. If everyone had acted honestly, perhaps search engines wouldn’t have had to become this smart.
The Great Leap: The Machine That Learned to Read Starts Writing The Large Language Models (LLMs) and Generative AI that amaze everyone today did not fall from the sky. The foundations of this technology were laid by the search engines’ 20-year struggle to “understand.” The process proceeded with a very clear logic:
- The Reading Phase: Search engines scanned trillions of documents to learn the structure, grammar, and semantic connections of human language. This was like a child learning to read.
- The Writing Phase: The system, having learned to read and understand at a perfect level, was no longer content with just finding and retrieving what existed. Using the patterns it learned, it began to seek the answer to the question, “What should the next word be?” In short, search engines were a massive school where artificial intelligence practiced “reading and comprehension.” The AI that learned to read in this school is now writing its own compositions.
History Repeats Itself: Tag Pollution on Social Media The “keyword fraud” experienced and solved on websites 20 years ago is now reappearing in a different form on social media. On platforms like Twitter (X) and Instagram, words that have become “Trending Topics” (TT) are embedded into posts that are 99% irrelevant to the content. A citizen clicking on a tag to get information about an earthquake encounters illegal betting ads or political propaganda. However, there is a tragic difference here compared to Google’s past situation: Google had to solve this problem because its reason for existence was “finding the right information.” Social media platforms, on the other hand, seem unwilling to solve this pollution. For them, what matters is not the accuracy of information, but the traffic and engagement created by that chaos. Although they technically have the power to filter out this irrelevant content in seconds, they prefer to commercially benefit from the “noise.”
Inertia in Software: Obsolete Checkboxes Despite this technological revolution, the Content Management Systems (CMS) we use (such as WordPress) still carry the ghosts of the past. The “tags” or “focus keyword” boxes that still sit under article panels create a fear of missing out (FOMO) in content creators: “If I don’t fill this in, my article won’t be found.” However, in the age of AI, it takes milliseconds for a machine to read a text and understand what it is about. Forcing humans to manually write tags is like trying to pull a modern car with a horse. Software interfaces need to shake off this inertia, hand over the classification task entirely to artificial intelligence, and leave humans free to focus solely on the “creative” part.
Conclusion The history of keywords is, in essence, the history of machines’ journey to understand human language. Improvements necessitated by fraudulent uses have acted as midwives to the birth of artificial intelligences that can converse with humanity today. Machines now know how to read, write, and understand. Our task is to move beyond old habits (manual tagging) and new tricks (hashtag spam) and use this technology to produce quality content.
A Note on Methods and Tools: All observations, ideas, and solution proposals in this study are the author’s own. AI was utilized as an information source for researching and compiling relevant topics strictly based on the author’s inquiries, requests, and directions; additionally, it provided writing assistance during the drafting process. (The research-based compilation and English writing process of this text were supported by AI as a specialized assistant.)
Aydın Tiryaki 