About

What this is

Cyberpolis is a research project that maps political positions. AI-generated arguments in the style of named public figures — politicians, pundits, and AI models themselves — are rated on a battery of stance axes and projected onto a two-axis map. AI models appear twice: as imputedpositions (what a generator model guesses they'd say) and as self-argued positions (what each model actually writes in its own voice, via its own API). The gap between those two channels — and where real users land when they answer the same questions — is the research. The full construction, reliability numbers, and caveats are on the methods page.

It is run as a personal research project, not a company. No advertising, no analytics tracker beyond what the host provides, and no plan to sell or share what you write. If that changes, this page changes first.

What this is not

• Not affiliated with any named public figure. When the site renders an argument "in the style of" a named politician, pundit, or other public figure, that is a public-facing label for a prompt instruction given to an LLM. The named figures have not been consulted, have not endorsed this product, and receive nothing. No copyrighted text from any figure is reproduced verbatim; the LLM's output is its own composition. If you are a named figure or representative and would like a name removed from the roster, contact the operator and it will be removed.
• Not affiliated with the model vendors. Anthropic, OpenAI, Google, xAI, Meta, Mistral, DeepSeek, Alibaba (Qwen), Moonshot (Kimi), Zhipu (GLM), and Baidu (ERNIE) are model providers accessed as third-party APIs. They have not endorsed this product, have not been consulted, and have no role in interpreting the results.
• Not a publisher. If you author an argument on this site, it is stored so the clustering map can include you, but Cyberpolis claims no ownership of what you write. You retain whatever copyright you hold over your own arguments.
• Not legal, medical, financial, or political advice. It is a research instrument for studying how LLM judges cluster relative to human political figures. The outputs are not endorsements of any position by anyone other than the human user, if a user authored the argument.

Contact

For takedown requests, name-removal requests, or any other operational matter, email jz@your-roost.com or visit jacobzucker.site.