Skip to content
Panelia
Public experiment

Élysée 2027

1,000 AIs asked about the 2027 French presidential election

This is not a tool, it is an experiment. We built a panel of 1,000 synthetic respondents and asked them about the 2027 French presidential election. The results are published in full.

Opens in a new tab, on its own domain.

This is not an opinion poll. The respondents are synthetic: they do not vote, they represent no real electorate, and these results predict no election. It is a method demonstration, published as it came out.

What it is for

To show, on a subject everyone knows, what a synthetic panel can and cannot do. The machinery is exposed: panel composition, raw answers, and the gap with what a polling institute would measure.

Who it is for

Anyone curious about what synthetic respondents are worth, and anyone who would rather judge Panelia's method on evidence than on a promise.

How it works

  1. 1

    Browse the panel composition and the method used.

  2. 2

    Read the results and the verbatims, exactly as they came out.

  3. 3

    Weigh them against your own expectations — and against real polls.

The other free tools

And when you want real answers?

These tools clear the ground. Panelia surveys hundreds of synthetic respondents and hands you a calibrated purchase-intent distribution, confidence intervals included.