For decades, digital presence was measured by one yardstick: if Google could find you, your company or your content, you existed online.

Artificial intelligence is quietly rewriting that rule.

As more people turn to AI assistants instead of search engines, a new question is taking shape: do AI models know who you are, and do they know you accurately?

That question is the premise behind In The Weights, an experimental platform that examines how people, organisations and brands are represented across multiple large language models. You can try it at intheweights.com.

How it works

Unlike a search engine, most large language models do not search the live web every time they answer a question. They draw primarily on knowledge encoded during training. In The Weights tests whether a person or organisation is represented in those model weights, and how consistently that representation holds across different systems.

The platform queries several leading models, including GPT, Claude, Gemini and Llama, and asks each to describe the same subject. It then compares the responses and produces a score estimating how consistently that subject appears across models.

That score is not a measure of reputation, influence or credibility. It is, at best, a rough estimate of representational consistency. Nothing more.

Why it matters

The score itself is the least interesting part of this exercise. The more important finding is structural: AI systems do not share a common understanding of the world.

The same person might be described accurately by one model, recognized only partially by a second and unknown entirely to a third. Even where a subject is recognized, the details can diverge sharply. Some models lean on stale training data, while others confidently produce details that are simply wrong.

None of this should be surprising. Each model is trained on a different dataset, built on a different architecture and shaped by different alignment choices. Each one ends up with its own internal, incomplete picture of people, organisations, products and events.

There is no single, shared AI memory.

A new layer of digital visibility

For years, organisations have invested in search engine optimization, media relations and reputation management. Those efforts were built around how people search the web.

AI assistants add a layer those efforts were never built for. Being visible on the open internet does not guarantee you are well represented inside a given model, and being well represented in one model offers no guarantee of the same in another.

As AI assistants become a more common way people discover information, how an organisation or individual is represented across models may become a meaningful part of digital strategy. Not a replacement for SEO, but an adjacent discipline with its own logic.

We are not there yet. But the trend is worth watching closely.

Final thought

In The Weights is best understood as an experiment, not a benchmark. Its methodology has real limitations, and its results should not be treated as definitive.

Even so, it points to a genuine shift. For years, the operative question was whether people could find you online. Increasingly, the question is whether AI models know who you are, and whether what they know is true.

That is a conversation worth having.

Ethics statement

This article examines a publicly available AI experiment and its broader implications for identity, visibility and model behaviour. It does not endorse the methodology, scoring system or conclusions produced by In The Weights. Readers should treat the platform as an experimental tool, not an authoritative measure of reputation, influence or factual knowledge.

Disclaimer

The views expressed are my own and do not reflect the views of my employer or any organisation with which I am affiliated. I have no financial relationship with In The Weights or its creators, and I was not compensated, sponsored or provided with any products, services or other consideration in connection with this article.

Keywords

#ArtificialIntelligence #AI #LargeLanguageModels #LLM #InTheWeights #GPT #Claude #Gemini #Llama #AIMemory #AIVisibility #AIIdentity #ModelWeights #GenerativeAI #DigitalIdentity #DigitalReputation