Thomas Dimson and Joey Flynn launched In the Weights, a site that queries major AI models—GPT, Claude, Gemini, Grok, and Llama—asking who someone is, then scores how well each model knows them from memory without web search. According to TechCrunch, the tool surfaces how embedded a person is in the training data these models rely on.
This looks like a vanity metric for the self-Googling set. People increasingly learn about founders, writers, executives, and public figures from chatbot answers that feel authoritative even when they're thin, outdated, or wrong. If you're not in the model, you're not in the conversation.
Publishers, marketers, and executives will scramble to become legible to AI systems that summarize the web instead of sending traffic to it. That fight centers on licensing deals, structured data, and whatever signals model makers decide to trust—a gatekeeping power held by OpenAI, Anthropic, Google, Meta, and xAI.
Google results used to be the scoreboard for public relevance. Now there's a second scoreboard inside systems you can't fully inspect or edit. In the Weights is more than a novelty. Status, discoverability, and credibility are being filtered through model memory instead of search rankings. The gatekeepers are the companies that trained the models.
Filed to the Technology desk · Jun 20