Forget Googling Yourself: ‘In the Weights’ Measures Your Digital Immortality via LLMs
Table of Contents
The New Metric for Digital Relevance
For two decades, the ‘vanity search’ has been a staple of the internet age: typing your own name into Google to see how the world perceives you. But as Large Language Models (LLMs) increasingly mediate how humans consume information, the search engine results page is no longer the final word on reputation. If a chatbot doesn’t know who you are, do you effectively exist in the eyes of the coming superintelligence?
This existential curiosity is the catalyst behind In the Weights, a new experimental platform designed to measure a person’s ‘weight’ within the neural networks of the world’s most powerful AI models. Created by Thomas Dimson and Joey Flynn, the tool shifts the focus from indexed web pages to the actual parameters—the weights—that define an AI’s internal knowledge base.
The premise is straightforward but technically revealing: the site determines if a model can recall a specific person without relying on real-time web browsing tools. In the world of AI, being ‘in the weights’ means your identity was significant enough during the training process to be encoded as a permanent part of the model’s memory, rather than being a piece of data fetched on the fly via a search plugin.
How the Scoring Works
To quantify this, In the Weights queries a diverse array of models—including GPT iterations, Claude, Gemini, Grok, and Llama—asking them to identify the individual. The system requests up to 10 results per model, complete with short descriptions and confidence intervals.
These responses are then clustered and analyzed to produce a ‘strength score.’ A high score indicates that multiple models recognize the person with high confidence and consistent detail. For instance, high-profile figures like Macaulay Culkin and Luciano Pavarotti frequently top the leaderboards, reflecting their deep embedding in the vast datasets used to train these models. However, the tool also reveals the inherent instability of AI memory. In some cases, models produce ‘hallucinations’ or struggle with ambiguity, such as misidentifying specific individuals as generic name forms.
From OpenAI to Existential Design
The project is the brainchild of Dimson and Flynn, both former OpenAI employees who joined the AI giant through the acquisition of their design startup, Global Illumination. According to Dimson, the project was born from a desire to explore how human lives are being translated into ‘floating point numbers inside the AI brain.’
Dimson noted that by 2026, the traditional Google vanity search will feel obsolete as more traffic migrates toward LLMs. The project draws inspiration from Terry Bisson’s short story ‘They’re Made Out of Meat,’ playing with the irony of biological entities seeking validation from silicon-based intelligence.
While some critics, including AI analyst Anthony Moser, argue that the tool is simply a streamlined way of asking thirteen different chatbots the same question, the psychological draw is evident. The platform’s retro, Nintendo-inspired aesthetic further gamifies the experience, turning the auditing of training data into a competitive leaderboard of digital permanence.
The Implications of Model Bias
Beyond the novelty, In the Weights highlights a critical gap in AI training: the discrepancy between who is ‘famous’ and who is ‘documented.’ Dimson intends to use the tool to analyze why certain models are biased toward specific demographics and to identify individuals who possess significant real-world influence—perhaps deserving of a Wikipedia page—yet remain invisible to the AI’s weights.
As the industry moves toward more specialized and curated training sets, the ability to be ‘remembered’ by a model may become a new form of social and professional currency. For now, In the Weights serves as a mirror, reflecting not how we are seen on the web, but how we are synthesized by the machines that are increasingly rewriting the internet.
