Two former OpenAI engineers have launched "In the Weights," a vanity-search tool that quantifies how strongly multiple AI models "remember" an individual. Built by Thomas Dimson and Joey Flynn, In the Weights takes a name and queries several large language models (LLMs) — including Grok, Gemini, various GPTs, Claude and Llama — with prompts such as "Who is <name>?", then clusters the responses to compute a proprietary metric called a "strength score." Each results page shows which models replied, the likelihood of hallucination, and an overall score.
Continue reading
The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in✓ Signed in — this article isn’t included in your current plan.