Highlight Twins validation study

In a blind study on concepts they'd never seen, Highlight Twins matched real Highlighters' answers with 92% of human-level consistency.

In this report

How Highlight Twins were tested against the real Highlighters they're modeled on
What 92% of human-level consistency means, and how it was measured
How Twins answered on product concepts they had never seen

Pulse runs on Highlight Twins: AI Twins, each modeled on one real Highlighter and built from millions of data points. To find out how well they hold up, we ran blind studies on product concepts the Twins had never seen and compared their answers with those of the real Highlighters they're grounded in.

The Twins matched nearly as often as people match their own earlier answers: 92% of human-level consistency. The full report covers how the studies were set up and what was measured.

Get the full report

Fill out the form below and we’ll send you straight to the download.