Assigning authority

I love having a research agent. Usually you have to be a tenured professor with a graduate assistant in order to get any interesting work done. The assistant does all the work – the professor claims the credit.

Right now I have a set of parallel subagents reading the chapters of a book passage by passage looking for changes in authority. I’ve been developing an idea that much of language exists for humans to make claims of authority over resources, each other, or even ideas. Who controls what? Thus I’m putting the AI to work testing my theory.

Think about how a king in ancient times would erect a stone tablet claiming to have built a temple. He didn’t build that temple. He just told a few advisors to build a temple, and they told a few guys with whips to build the temple, and they, finally, told the actual people who built the temple to build the temple. You couldn’t convince anyone the king did anything worthwhile if you didn’t have words.