Entity, Not Ego: How AI Sees You as a Thing, Not a Person

Your ego thinks you're a person with a story. To an AI, you're a thing, an entity it has to recognise, connect, and decide whether to trust. Once that clicks, a lot of confusing stuff, why the wrong person keeps getting recommended, why a big personality sometimes gets overlooked, starts making a lot more sense.

What an entity actually is

To a machine, you're a node. A distinct thing with a name, some attributes, a body of work, and connections to other things. It builds this picture from everywhere it finds you and tries to decide: is all of this the same person, and can I trust what's attached to them? That node, not your personality, is what gets recommended or skipped.

Entity resolution, explained the way a friend would explain it

There's a bit of jargon worth demystifying here, because the concept underneath it is actually simple. Entity resolution is the process a system uses to decide which mentions scattered across the web all refer to the same real person. Picture the everyday version of this: you mention "John Smith" to a friend, and before they respond, they pause and ask, wait, which John Smith do you mean, the one from accounting or the one who plays in your football league. That pause is entity resolution happening in a human brain, weighing context clues to figure out which real person is being discussed. An AI system does a version of the exact same thing, except instead of a quick mental check, it's sifting through pages, profiles, and mentions trying to decide whether they cluster into one coherent person or several different ones. Disambiguation is the closely related term for the part where it actively tells two similar candidates apart, using whatever distinguishing details it can find, a specific employer, a city, a niche, a photo, a consistent bio.

The knowledge graph analogy, in plain terms

If you've ever looked up a public figure and seen a neat box of facts appear alongside the search results, birthplace, occupation, notable works, that box is usually pulled from something called a knowledge graph, and Wikidata is one of the best known open examples of one. In a system like that, a person isn't stored as a warm paragraph about their personality or their journey, they're stored as a node with a set of attributes (born here, works there, known for this) and a set of relationships to other nodes (employed by this organization, co-authored with that person, alumnus of this school). It's a structural, almost clerical way of representing a human being. AI systems reason about you in a spiritually similar way even when no formal Wikidata entry exists for you personally, they're constantly trying to assemble that same kind of structural picture from whatever scattered facts they can find, then deciding how confident they are in it.

Figure: the rough sequence a system works through before it's confident enough about who you are to recommend you by name.

Why the machine is unimpressed

Here's the thing. People pour energy into being impressive to humans and forget to be legible to machines. But if the engine can't confidently resolve who you are, it won't risk naming you. Being recognised as a clear entity comes before being recommended. You can't win a game the machine doesn't know you're in.

A hypothetical: two therapists named Alex Rao

Imagine there are two licensed therapists, both named Alex Rao, practicing in different states, one specializing in trauma work, the other in couples counseling. Neither has anything to do with the other, but from an AI system's point of view, that's not obvious at first glance, it just sees the same name attached to two different sets of facts. If both Alex Raos have thin, inconsistent bios, the system may genuinely struggle to keep them apart, occasionally blending a fact from one into an answer about the other, or simply refusing to name either one confidently because the picture is muddy. The trauma specialist Alex Rao who wants to be reliably recommended needs to make disambiguation easy for the machine: a full middle name or credential included consistently, a specific city and license number where appropriate, a niche described in the same words everywhere, so that anyone, human or machine, encountering the name knows immediately which Alex Rao is being discussed. This is exactly the kind of entity confusion covered from a different angle in AI got the facts wrong about me, since a same-name mixup is one of the most common reasons facts get scrambled in the first place.

Attributes and relationships matter more than adjectives

One subtle shift worth internalizing: a knowledge graph style entry almost never contains adjectives about how good someone is at their work. It doesn't say charismatic, passionate, or visionary, it says facts, is employed by, has written, is credentialed in, is located in. AI systems lean heavily on that same kind of factual, relationship-based information when deciding what they can confidently say about you. A glowing description of your own brilliance, written by you, carries far less weight than a boring, verifiable fact repeated consistently across several independent sources. This is uncomfortable for anyone who has spent years crafting a compelling personal narrative, because it means the narrative isn't what's being read most closely, the underlying scaffolding of facts and relationships is.

The fragmentation problem

Most professionals accidentally exist as several entities. Your full name on one platform, a nickname on another, a different job title here, an outdated bio there. To you, obviously one person. To an engine, possibly three weakly-connected profiles, each holding a fraction of your signals, none strong enough to name. Your reputation is real. It's just scattered. If you want to see exactly how scattered, the audit described in googling yourself versus asking ChatGPT is a quick way to spot the gap between what a traditional search engine shows and what a conversational AI actually pieces together about you.

How to become one clean entity

Consolidate. One canonical name, spelled the same everywhere. One consistent title and bio. Properties that link to each other so the machine can trace them to a single you. Where possible, structured data that explicitly says "this name, this bio, these profiles, this work, all one entity." Boring, unglamorous work. Decisive, compounding results over time. It also helps to understand which sources an AI system tends to trust most when it's trying to resolve who you are in the first place, since a mention on a highly trusted source does more disambiguation work than ten mentions on obscure ones, a subject covered fully in Wikipedia, Wikidata and why machines trust them.

The payoff of being legible

When your entity is clean, every article you publish and every mention you earn compounds onto one recognisable node instead of splitting three ways. The engine gets more confident with each signal, because it knows exactly where to attach it. That confidence is what tips a maybe into a named recommendation. It's also worth remembering that general fame and being a clearly resolved entity are not the same thing either, a widely known name with a messy, contradictory footprint can be harder for a machine to confidently vouch for than a lesser known name with a clean one, a nuance explored in unknown versus celebrity in AI answers.

A short exercise: describe yourself like a knowledge graph would

Here's a useful, slightly humbling exercise. Try writing a description of yourself using only the kind of language a knowledge graph would use, no adjectives, no narrative, just verifiable attributes and relationships. Something like, works as a [role] at [organization], based in [location], holds [credential], has published [specific, named piece of work], has been referenced by [specific outlet or person]. If that exercise leaves you with a short, thin list, that's not a personality problem, it's a signal that your public footprint hasn't yet given a machine much structural material to work with, regardless of how compelling your personal story is in a conversation. If the list comes out long and specific, you're probably already closer to being a well-resolved entity than you think, and the next step is simply making sure that same list of facts appears consistently everywhere your name shows up.

Check your ego at the schema

None of this means your story, voice, and personality don't matter, they're what make humans choose you once you're found. But the machine that decides whether they find you doesn't care how you feel about yourself. It cares whether it can resolve and trust your entity. Serve that first, patiently and a little unglamorously, and your ego can absolutely take the win later, once the machine has actually decided it can vouch for you.