As traditional web search engines face increasing scrutiny over their relevance, a new platform called In the Weights is shifting the focus toward how individuals are represented within the internal parameters of artificial intelligence. Created by Thomas Dimson and Joey Flynn, the site aims to quantify an individual’s presence by measuring how well various large language models (LLMs) can recall specific people without relying on live web search tools.

The platform functions by querying a diverse array of AI models—including GPT, Claude, Gemini, Llama, and Grok—with requests for information about specific names. It then clusters the resulting descriptions and assigns a ‘strength score’ based on the model’s confidence and consistency. The site suggests that being ‘in the weights’ signifies that an individual’s identity has been encoded into the foundational training data of these systems, effectively serving as a modern vanity metric for the AI era.

Early results have highlighted the competitive nature of this digital recognition, with celebrities and public figures currently dominating the leaderboard. Beyond the novelty of the scores, the project provides insights into how different models hallucinate or struggle with ambiguity, offering a unique window into the ‘brains’ of current AI technology. The creators view this as a necessary evolution in how we perceive personal relevance, moving away from search engine rankings toward a model-centric understanding of digital existence.

Looking ahead, the developers plan to expand the platform’s capabilities to analyze model biases and identify individuals who lack traditional documentation, such as Wikipedia entries, but remain prominent within AI training sets. While some critics dismiss the tool as a simple aggregation of chatbot responses, the project has gained significant traction, tapping into a growing public curiosity about how human identity is being preserved or distorted within the next generation of super-intelligence.

Key Takeaways

• In the Weights measures how well AI models recall individuals based on their internal training data rather than live web searches.
• The platform assigns a 'strength score' to users by aggregating responses from multiple LLMs, including GPT, Claude, and Llama.
• The project aims to explore how human identity is encoded in AI, highlighting potential hallucinations and model biases in the process.

Editor’s Analysis & Impact

The emergence of ‘In the Weights’ signals a broader cultural shift in how we define digital relevance. As LLMs become the primary interface for information retrieval, the traditional SEO-driven vanity search is losing its status as the ultimate arbiter of fame. This tool highlights a critical transition: the move from ‘being found’ on the web to ‘being known’ by an AI. From an industry perspective, this underscores the growing importance of training data composition. If an individual’s presence in an AI model becomes a status symbol, we may see a future where ‘AI optimization’ becomes as vital as search engine optimization. However, the project also exposes the inherent instability of these models, as varying scores across different versions of the same AI demonstrate that ‘digital immortality’ is currently subject to the whims of training methodologies and algorithmic hallucinations.

Frequently Asked Questions

Q: How does In the Weights calculate its score?
A: The platform queries multiple AI models to describe a specific person, clusters those descriptions for similarity, and assigns a strength score based on the models' confidence and consistency in their responses.

Q: Does the site use live web search to find information?
A: No, the site specifically tests the models' internal knowledge—the 'weights'—to see if the information is encoded within the AI's training data without the use of external search tools.