AI Vanity Search Gets a Scoreboard With In the Weights

Googling yourself used to be the easiest way to see how the internet remembered you. Now, a new site called In the Weights is asking a more AI-era question: do large language models remember you?

Created by Thomas Dimson and Joey Flynn, In the Weights measures how well different AI models can recall a person’s name without using web search. The site then turns those model responses into a “strength score,” creating a kind of AI-powered vanity search leaderboard.

The Quick Read

The New Vanity Search

In the Weights is built around a simple but strange idea.

Instead of searching Google, the site asks multiple AI models a question like: “Who is this person?” It then compares how often and how confidently those models recognize the name.

The “weights” refer to the numerical parameters inside AI models that shape how they respond. The site says it is trying to measure whether a person has been encoded inside those model weights, rather than simply found through live search.

That turns personal online visibility into something different. It is no longer only about ranking on Google. It is also about whether AI systems have absorbed enough information about you to describe you from memory.

How the Score Works

The site queries several models, including versions of GPT, Claude, Gemini, Grok, Llama and smaller AI systems.

It asks each model to return possible identities for a name, with short descriptions and confidence levels. Then it groups similar descriptions together and assigns a strength score based on how strongly the models appear to recognize the person.

That score creates the addictive part. Users can compare themselves with writers, founders, celebrities, public figures and friends.

TechCrunch’s Anthony Ha reported that his own result gave him a score of 641, putting him in the top 6% of names checked at the time. The leaderboard was shifting during his test, with Macaulay Culkin and Luciano Pavarotti near the top.

Built by Former OpenAI Employees

Dimson and Flynn previously worked at OpenAI after their design startup Global Illumination was acquired.

Dimson told TechCrunch the idea came from the feeling that Google vanity searches are becoming less relevant as more people learn about others through AI chatbots. He described the project as a way to explore how many lives are now “encoded” inside AI model parameters.

That makes In the Weights feel playful, but also slightly uncomfortable. It is both a joke about internet ego and a serious signal about how identity is moving into AI systems.

Hallucinations Are Part of the Fun

The site also shows when models appear to get things wrong.

For example, TechCrunch noted that one model described “Anthony Ha” as an ambiguous name form related to initials, rather than identifying the journalist correctly.

That matters because AI recall is not the same as truth. A model may confidently describe the wrong person, combine details from multiple people, or invent a plausible-sounding identity.

So a high score may feel flattering, but it does not necessarily mean the model understands you accurately.

A Reputation Tool or Just an Internet Game?

In the Weights sits somewhere between a novelty site and a glimpse of the future.

On one side, it is clearly designed to be fun. It has a retro-inspired look, scores, rankings and the competitive pull of seeing who is more “remembered” by AI.

On the other side, it points to a real shift in online reputation. As AI assistants become a bigger source of answers, people and brands may start caring not only about search results, but also about how AI models describe them.

That could eventually create a new type of reputation management: not search engine optimization, but model memory optimization.

The Bigger AI Identity Question

The site raises a bigger question: what does it mean to exist inside an AI model?

If a model can describe you without live search, it likely absorbed public information about you during training. But users usually cannot see exactly where that information came from, how accurate it is, or how strongly it shaped the model.

That creates a strange new layer of identity. Your public online footprint may not only appear in search results. It may also live inside AI systems as statistical memory.

In the Weights turns that abstract idea into a score people can actually see.

Final Take

In the Weights is a clever, slightly unsettling product for the AI age.

It makes vanity search feel new again by moving it from Google results to model memory. The site is fun because it gives people a score, but it is interesting because it reveals how much online identity is changing.

As more users ask chatbots about people, companies and creators, the next reputation question may not be “What shows up when I Google myself?”

It may be: “What do the models think they know about me?”