Exactly. So, I think augmenting your boss with some… on OpenAI, they’re calling it collective alignment. So basically, ways to facilitate conversations in the population. We work with OpenAI on that - sending surveys to 1,000 statistically representative Americans and asking them not to answer predefined questions, but type your own statements on how AI should behave, and then rank those preferences from each participant to each statement. So, you have this giant matrix of social preferences.