Googling yourself is out; prompting yourself is in. Former OpenAI engineers Thomas Dimson and Joey Flynn just launched In the Weights, a new platform designed to measure your AI-centric vanity score. According to TechCrunch AI, the tool checks multiple large language models to see if your identity has been permanently encoded into their training data, or weights.
This is a fascinating development because it highlights a major shift in how we seek and verify information. For decades, a Google search was the canonical proof of relevance. Now, as chatbots increasingly become the default arbiters of knowledge, Dimson and Flynn recognized that traditional web searches are becoming an outdated metric. Being in the weights means your existence was deemed important enough to be baked into the foundational parameters of a model, entirely independent of live web search tools.
Here is a breakdown of what In the Weights brings to the table and how it functions:
- The multi-model scoring engine: Instead of relying on a single AI, the platform queries a wide spectrum of models including Grok, Gemini, Claude, Llama, and multiple versions of OpenAI’s GPT. It asks a simple question: “Who is [name]?” and requests up to 10 results with short descriptions and confidence levels. The system then clusters similar descriptions and assigns a numerical “strength score” based on how well the models recall the person.
- Exposing AI hallucinations: Beyond just serving as an ego boost, the tool provides a transparent look at how different models process the exact same prompt. It shows specifically which models returned which answers for a given name, making it remarkably easy to spot hallucinations. For instance, TechCrunch AI reports that one model completely hallucinated a response for their writer, categorizing his name as an “ambiguous name form.”
- The global vanity leaderboard: To gamify the experience, In the Weights features a real-time leaderboard. As of the initial launch, celebrities like Macaulay Culkin and Luciano Pavarotti were battling for the top slot with scores nearing 1,000. It effectively turns AI model recall into a competitive, highly shareable social metric.
- A retro, accessible interface: The platform is currently live and free to use. It wraps this complex multi-model querying system in a charming, Nintendo-inspired retro design. This aesthetic choice makes an otherwise deeply technical concept, comparing the floating-point numbers inside various AI brains, highly approachable for the general public.
While the reception has been massive, the platform does have its detractors. Not everyone believes this represents a profound technological breakthrough. As detailed in the original report, AI critic Anthony Moser dismissed the platform, pointing out that it is “literally the same as asking 13 chatbots to tell you about yourself.”
Despite the criticism, the creators have ambitious plans for the underlying data. Dimson intends to investigate why different models within the same series return conflicting results, analyze model biases toward specific demographics, and even identify notable individuals who lack Wikipedia pages but are highly recognized by artificial intelligence.
Whether you view In the Weights as a true measure of digital immortality or simply a clever party trick, it successfully taps into our collective curiosity about what these systems actually know about us. You can find more details and context at the original source.