In the Weights Debuts as a Metric for AI Recognition of Individuals

What happened

In the Weights is a new AI vanity search tool built to answer an unusual question: how strongly do large language models recognize a specific person without using live web search? Created by former OpenAI designers Thomas Dimson and Joey Flynn, the project tests whether a person's identity is sufficiently embedded in model parameters, or "weights," to be recalled directly from training rather than fetched from the internet in real time.

That framing makes the product feel novel even in a crowded AI-tools market. Traditional vanity search has long meant looking yourself up on Google. In the Weights updates that idea for the LLM era by asking not whether the web contains information about you, but whether major models already "carry" a stable version of who you are.

What's new in this update

The major development is the launch of a public-facing system that converts that abstract idea into a measurable score. In the Weights queries models such as GPT, Claude, Gemini, Grok, and Llama, then compares how consistently and confidently those models describe a given individual. The result is a kind of AI recognition score that aims to capture how visible a person is inside model training data and internal model memory.

That matters because AI products are increasingly replacing ordinary search behavior. As people ask chatbots for summaries, biographies, and recommendations, presence in model knowledge may become a new kind of digital status marker separate from classic SEO or social-media reach.

Key details

The tool works by prompting multiple models to identify or describe a person without access to web retrieval. It then clusters and evaluates the outputs to determine whether the person is clearly recognized, inconsistently recognized, or misidentified. That creates a leaderboard-style experience, but the underlying concept touches on deeper questions about memory, visibility, and AI-era reputation.

Several features make the project notable:

• It tests model recognition without live web search
• It produces a "strength score" for identity recall
• It surfaces hallucinations and ambiguity across model families
• It compares how different LLMs encode public figures

The tool reportedly highlights cases where one model recognizes someone cleanly while another returns an uncertain or incorrect answer. That difference is part of the appeal. It turns AI inconsistency into an observable product rather than a hidden system quirk.

Background and context

The creators, Dimson and Flynn, previously worked in the OpenAI orbit after the acquisition of Global Illumination, and their project taps directly into a growing cultural shift. People are beginning to wonder not only what AI can do, but how AI "sees" them. For public figures, technologists, and creators, that question can feel like a new layer of prestige. For ordinary users, it can feel like an eerie mirror of digital relevance.

That is why the phrase digital immortality attached to the product resonates. The site is playful, but it also points to something real: large language models are becoming another archive layer in how identity is stored, summarized, and presented. If a model can describe you without searching, then in a limited sense you already exist inside its learned representation of the world.

At the same time, the tool also exposes a warning. Recognition is not the same as accuracy. A person can be "in the weights" and still be distorted, flattened, or confused with someone else. That tension makes the project more than a gimmick. It becomes a visible demonstration of both AI memory and AI error.

What to watch next

The next interesting question is whether In the Weights becomes a curiosity or the start of a broader category around AI identity diagnostics. As more people rely on chatbots for biography, expertise lookup, and reputation framing, tools that measure LLM visibility could become more common.

Three follow-up directions are especially worth watching:

• Whether the creators expand scoring methods or model coverage
• Whether users begin treating AI recognition as a reputational benchmark
• Whether the tool reveals systematic bias in who models remember well

If those questions deepen, the project could evolve from novelty search into a serious commentary on visibility in the AI era.

Why this matters

The In the Weights AI vanity search matters because it translates an abstract machine-learning concept into a human question people instantly understand: do leading AI models know who you are? That makes the tool culturally sticky, but it also highlights how much influence LLMs may have over future discovery, reputation, and public memory.

More broadly, the project is a reminder that as AI becomes a search layer, being on the web may no longer be the only measure of digital presence. Being in the weights may become its own strange form of relevance.