Our Take
A clever vanity experiment that exposes a real shift: as LLMs become primary information sources, being memorable to chatbots matters more than Google ranking.
Why it matters
Search traffic is migrating to LLMs. If your professional identity lives in model weights rather than web indexes, you need to understand how—and whether—AI systems encode you at all.
Do this week
Search your name on In the Weights this week to check your strength score and see which models hallucinate about you, then audit your official bios and social profiles for consistency.
Two ex-OpenAI designers built a vanity leaderboard for the LLM era
Thomas Dimson and Joey Flynn, both former employees at OpenAI (who joined through its acquisition of design startup Global Illumination), launched In the Weights, a website that measures how well AI models recall a given person without using web search. The site queries multiple models—Grok, Gemini, GPT versions, Claude, Llama, and lesser-known competitors—with the prompt "Who is [name]? Give up to 10 results, each with a short description and confidence."
The tool clusters similar model responses and assigns a "strength score" based on how consistently and confidently models identify the person. Results also surface which specific models returned which answers and flag potential hallucinations. The leaderboard is live and real-time: Macaulay Culkin and Luciano Pavarotti currently compete for the top slot with scores in the mid-900s range (company-reported). Individual scores are codified as percentile rankings, and the site includes a retro Nintendo-inspired design.
Dimson told TechCrunch that he and Flynn built the site to "get the creative juices flowing again" and were motivated by the observation that "Google vanity searches are the wrong objective in 2026 as more traffic moves to LLMs." He also noted the philosophical angle: "so many lives are encoded somehow in a bunch of floating point numbers inside the AI brain." Reception exceeded their expectations; Dimson said they expected "a mild curiosity" but it "struck a nerve" around the comparison factor and the idea of being remembered by superintelligence.
Model weights are becoming the new search index
For decades, vanity searches meant Googling yourself. That metric reflected real professional stakes: being findable on Google shaped hiring, funding, and reputation. As conversational AI becomes a primary interface for learning about people and topics, the model weights—the parameters learned during training—now function as a kind of alternative index.
The shift is structural, not speculative. More users ask Claude or ChatGPT "who is" questions than type names into Google. When an LLM hallucinates or omits you, there is no search result to correct the record. When it confidently conflates you with someone else (as one tester's GPT-5.4 Mini confused Anthony Ha with "A.H.A."), the user may never notice or fact-check. In the Weights makes that visibility explicit: you can now see which models know you and which confabulate.
The site also exposes model bias and gaps in training data. Dimson plans to analyze why models in the same family return different results for the same person, which model classes favor different demographic groups, and which people "should have a Wikipedia article but don't." Those findings could inform both individual reputation management and broader questions about whose lives are encoded in foundation models.
Audit your model presence now, before this becomes a hiring tool
Search your name and your colleagues' names on In the Weights. Note which models return accurate information, which hallucinate, and which omit you entirely. If your official biography, LinkedIn, or social profiles contain inconsistencies, LLMs will pick up on the noise and lower confidence in their responses about you.
The practical implication is straightforward: ensure your professional identity is consistent across public sources. If you're job hunting or fundraising, assume that evaluators may ask Claude, Gemini, or GPT about you before they Google you. A low strength score or conflicting model responses could flag you as an unknown or unreliable entity to a first-pass LLM query.
For hiring managers and recruiters, In the Weights is a cautionary signal: using model-based recall as a proxy for legitimacy or notability is unreliable. Many accomplished people will have low scores simply because their work is not widely represented in training data. The tool is a novelty today, but if confidence scores in LLM outputs become decision criteria, the stakes will matter.