In the Weights: AI vanity search that scores your name

In the Weights is a new website built by Thomas Dimson and Joey Flynn that measures how well large language models can recall a person without using web search. The site queries multiple models, clusters similar descriptions and assigns each name a numeric strength score to show who is “in the weights.”

What does In the Weights do and how?

In the Weights queries several models, including Grok, Gemini, multiple versions of GPT, Claude, and Llama, with a prompt like “Who is ? Give up to 10 results, each with a short description and confidence,” clusters similar descriptions and assigns a strength score. The site then displays which models returned which answers and highlights possible hallucinations. That clustering-and-scoring process is presented as a measure of how much a person’s existence is encoded in a model’s parameters, the creators say.

The project’s founders built the site after leaving OpenAI through the acquisition of their design startup Global Illumination. Dimson framed the idea around a shift away from traditional Google vanity searches, saying that “Google vanity searches are the wrong objective in 2026 as more traffic moves to LLMs” and that many lives are “encoded somehow in a bunch of floating point numbers inside the AI brain.”

Who ranks highest and what do the scores mean?

Scores are single numeric values meant to indicate how strongly a name is represented across models; one example in the site shows a strength score of 641, placing that person in the top 6% of names, while Macaulay Culkin currently sits in the top slot with a strength score of 988, neck-and-neck with Luciano Pavarotti. The site’s leaderboard updates as models return different outputs.

The site also surfaces model-specific quirks. For example, GPT-5.4 Mini returned a suspicious answer about the author Anthony Ha, calling him “an ambiguous name form that could refer to multiple people with the initials A.H.A.,” which the site flags as a potential hallucination. The leaderboard and per-model answer breakdown let users compare which models recall which people and where disagreements or errors appear.

How did the creators position the project?

Dimson said the project began as a creative experiment after he and Flynn left OpenAI, and that an offhand blog post riffing on AI weights and Terry Bisson’s short story “They’re Made Out of Meat” helped seal the site’s direction. He described early uptake plainly: “Reception has been insane so far, we thought this would be a mild curiosity but it seems like it has struck a nerve of wanting to see if you live forever in the super intelligence (the comparison factor doesn’t hurt either!).”

The site pairs its scoring with a playful, retro Nintendo-inspired design and explicitly calls out when models likely hallucinate information. Critics have been blunt: one AI critic summarized the idea as “literally the same as asking 13 chatbots to tell you about yourself.”

Why it matters

In the Weights turns a technical property of models into a public metric people can compare. As the creators put it, the project reframes the vanity search for an era when more people interact with LLMs than with traditional search. The site exposes differences in model training and recall, making visible which models encode which names and where hallucinations or biases appear.

This matters because the measurement treats model recall as social signal. Names with high strength scores are easier for multiple models to retrieve without tools, and that visibility can reinforce online reputations or highlight omissions where someone “should have a Wikipedia article but don’t,” a research angle the founders plan to explore.

What to watch

Dimson plans to dig into why different models in the same series return different results, which models skew toward particular person types, and which people appear to be omitted from model training data. Expect deeper comparisons of model bias, per-model recall patterns and more scrutiny of hallucination cases.