Testing the portrait

How "Richie" is RichieBot?

We are using validated psychological assessments to measure how closely RichieBot’s personality matches the Richie his friends and family knew and loved.

Friends and family informant reports
12
Validated psychological assessments
3

A shared test

Three conditions, assessed with psychological measures

Friends and family fill out psychological assessments about Richie. Both RichieBot and a naïve LLM with no information about Richie complete the same measures. We examine the correspondence between each instance of measurement.

RichieBot, friends and family, and a blank control converge into the same Big Five, Schwartz values, and primal world beliefs assessments, which lead to an overlapping profile shape and two statistical comparison questions.

What counts as similar?

For every survey measure, we ask two questions.

01

Similarity to human raters

We calculate a 95% confidence interval which indicates an acceptable margin of error around the aggregate friends-and-family rating then test whether RichieBot’s score falls within that interval.

02

Closer than the blank control

We compare the distance between each model’s score and the ground truth, then test whether RichieBot is closer than the uninformed control persona.

Results

Assessment-by-assessment findings

Results will be added here when the analysis is complete.

01

Big Five Personality Traits

Personality traits

Results forthcoming

02

Primal World Beliefs

Beliefs about the world

Results forthcoming

03

Schwartz Values

Personal values

Results forthcoming