Are You 'In the Weights'? The New AI Vanity Search Replacing Google

For more than two decades, the "vanity search"—the act of typing your own name into Google to see exactly what the internet thinks of you—has served as a universal digital habit. It was the simplest way to measure your digital footprint, check your professional visibility, or simply ensure that no embarrassing legacy content was floating at the top of the search results. But as the architecture of the web undergoes a fundamental evolution, the canonical source of human information is rapidly shifting. With millions of users now bypassing traditional search engines entirely in favor of conversational interfaces, the old metrics of online relevance are beginning to look obsolete. The question is no longer what Google's index says about you, but rather what the next generation of artificial intelligence inherently knows about you.[1][5]

As large language models (LLMs) like ChatGPT, Claude, and Gemini increasingly become the default tools for research, discovery, and daily inquiries, a new frontier of digital identity has emerged. When a user asks an AI chatbot about a public figure, a colleague, or a brand, the system doesn't always scour the live web for an answer. Instead, it often relies entirely on its internal memory—the vast, compressed ocean of data it absorbed during its initial training phase. This shift has prompted a fascinating, slightly existential question for anyone who lives and works online: does artificial intelligence actually know who you are, or are you entirely invisible to the machine?[5]

Enter "In the Weights," a newly launched and highly viral website that serves as the internet's first dedicated AI-centric vanity search. Created by Thomas Dimson and Joey Flynn—two former OpenAI employees who joined the AI giant through the acquisition of their design startup, Global Illumination—the tool is designed to measure exactly how deeply a person is embedded in the internal memory of the world's leading artificial intelligence models. Rather than counting backlinks or keyword density, the platform provides a standardized metric for AI recall, offering users a unique, quantifiable glimpse into their own digital legacy.[1][3][5]

The project's name is a direct reference to the underlying architecture of modern neural networks. In machine learning, "weights" are the billions of numerical parameters that form an AI model's brain. During the training process, as a model ingests petabytes of text from across the internet, it adjusts these weights to encode knowledge, relationships, and facts. If a model can accurately recall your biography, your achievements, or your profession without relying on a live web search, it means your digital footprint was significant enough to permanently alter those numerical parameters. As the creators explain, being "in the weights" means your existence was deemed important enough to be baked into the foundational mathematics of superhuman technology.[3][5][6]

To calculate this modern metric of relevance, In the Weights operates by simultaneously querying a gauntlet of the industry's most powerful models. The system sends prompts to frontier models like OpenAI's GPT-5.5, Anthropic's Claude Opus 4.8, and Google's Gemini, while also testing xAI's Grok and various open-weight systems like Meta's Llama series. The tool deliberately restricts these models from accessing the live internet, forcing them to rely entirely on their pre-trained internal memory to answer a simple question: "Who is this person?"[2][4][5]

The methodology behind the tool is both clever and rigorous. When prompted with a name, the system demands that each AI model return up to ten distinct results, complete with short biographical descriptions and a self-reported confidence level. Because many people share the same name, the platform then utilizes a clustering algorithm to group similar descriptions together, filtering out the noise and isolating the specific individual the user is actually searching for. By aggregating how many models successfully recall the person and how confident they are in their answers, the system generates a unified metric.[2][5]

This aggregated data is distilled into a final "strength score" that ranges from zero to a hard, mathematically defined ceiling of 996. According to Dimson and Flynn, that maximum score is practically unattainable for a standard internet user; it is strictly reserved for historically monumental or universally recognized names whose data saturates every corner of the internet—figures like William Shakespeare, Wolfgang Amadeus Mozart, and modern pop icons like Taylor Swift. The calibration of this scale is itself a statement about what the tool is measuring: it is not just looking for a passing mention, but for profound, inescapable cultural saturation.[3][4]

Currently dominating the modern celebrity leaderboard is actor Macaulay Culkin, who boasts a staggering strength score of 988, a testament to the sheer volume of cultural references, trivia, and pop-culture data surrounding the *Home Alone* star. He is closely followed by the late, legendary opera singer Luciano Pavarotti. For everyday users, founders, and journalists testing the system, the numbers are far more grounded. Hitting a score of 641 is enough to place an individual in the top six percent of all searched names, though the leaderboard shifts constantly as more users input their data and test the boundaries of the system.[1][2][5]

Beyond the sheer novelty of chasing a high score, In the Weights functions as a fascinating, publicly accessible diagnostic instrument for AI behavior. For researchers and developers, the tool provides a rare, comparative look at how different model architectures handle data compression and factual recall. One of the clearest practical findings from the tool's early usage concerns the performance of smaller, open-weight models compared to their massive, frontier-scale counterparts.[4]

The creators note that appearing in a smaller model—such as Meta's highly efficient one-billion-parameter version of Llama—is actually a much stronger indicator of cultural relevance than appearing in a massive model like GPT-5.5. Because smaller models have strictly limited parameter counts, they are forced to compress human knowledge far more aggressively during training. They simply do not have the space to remember niche internet micro-celebrities or obscure professionals. Therefore, if a small model retains a specific person's biography, it means that individual survived a rigorous, algorithmic culling process that only preserves the most highly repeated data.[3][4]

The platform also serves as a real-time, interactive monitor for AI "hallucinations"—the well-documented phenomenon where a model confidently invents false information when it lacks sufficient factual data. When querying names that fall into the long tail of internet obscurity, the models often struggle. They might invent entirely fictional biographical details, confidently confuse an ordinary user with a historical figure who shares the same name, or merge the achievements of two different professionals into a single, hallucinated entity.[2][3][5]

In the Weights transparently flags these discrepancies for the user, highlighting exactly where a model's genuine memory ends and its creative fabrication begins. This feature is particularly illuminating because it demonstrates that a high strength score is not necessarily evidence that the model's account of you is perfectly accurate—only that the model is highly confident in its internal representation of you. It is a stark reminder that as we increasingly rely on AI for information, the "facts" it presents are often probabilistic guesses rather than hard database lookups.[2][5]

Wrapped in a nostalgic, retro Nintendo-style pixel-art interface, the tool playfully masks a profound and permanent shift in how digital legacy is measured and managed in 2026. For professionals whose careers and reach depend heavily on discoverability—such as academic researchers, startup founders, specialized journalists, and digital creators—being "in the weights" is rapidly replacing traditional Search Engine Optimization as the ultimate metric of online success.[1][2][4]

In the previous era of the web, digital strategists focused entirely on optimizing websites to rank highly on Google, ensuring that a vanity search yielded pristine, controlled results. But as Dimson noted in a recent interview discussing the project, searching for oneself via Google is steadily losing its significance. Today, he argues, "many people's lives are encoded within the decimals in the AI's brain." If an executive or brand isn't accurately recalled by the foundational models powering the next generation of chatbots, they risk being entirely invisible to a massive segment of the internet public.[2][4][6]

Ultimately, In the Weights provides a unique, highly engaging mirror for the generative age. It takes the highly technical, often opaque concept of machine learning parameters and translates it into a deeply human, accessible experience. As artificial intelligence continues to evolve from a novel tool into the primary infrastructure of global knowledge, this platform answers a question that is equal parts fascinating and unsettling: in the process of building superhuman technology, did the machine deem your existence important enough to remember?[5][6]