Thread by @alexpriou

A.I. and The Humanities—An Experiment

I plan to spend the Fall studying Machiavelli's Discourses on Livy, and for that I will use Claude liberally.

The goal: to see how much it can accelerate the process of close reading.

Call it "Artificial Thoughts on Machiavelli."

🧵

I'm going to use it in a few ways.

1) Collect sources

Machiavelli draws on a number of different texts as he proceeds.

As I go chapter-by-chapter, I will have Claude comb the web for sources people have identified, and compile them into a single document for my review

2) Analyze discrepancies

Machiavelli infamously departs from the sources he draws on, often with intended meaning.

So, I will have Claude add to the compilation of sources a preliminary list of discrepancies, again for my review.

3) Transcribe my dictated thoughts

After a review of the text and Claude's material, and anything else I am reading, I will dictate my thoughts on the chapter for Claude, or another A.I., to transcribe.

I will edit for accuracy, then give it Claude for review.

4) Refinement of the Strauss-bot

I will ask Claude for feedback, and correct Claude/my work as needed. I will ask it to apply these results going forward.

That last step will be the real test. Can it learn how to read, to really read? Or will it fall into typical ruts of thinking?

Can it actually notice things, can it develop an eye, or at least more of an eye than we Humanities professors give it credit for?

I want to take this new tool seriously—it has already been very helpful with mundane tasks.

I want to give it the benefit of the doubt, so I am going to read a very difficult text with it, and see what it can do.

It certainly helps with the grunt work of collecting primary sources.

It has helped, too, with finding passages for me in primary sources, when my memory needed jogging.

It clearly speeds things up. Not without a cost, I'm aware, but it's not a total loss—it's a trade-off.

I'm already aware of the benefits of the old school way. But how can one assess trade-off, without assessing the benefits of the new school way?

I'll report back when I have some conclusions I'm confident in.