AI Experiments Steven 2: Searching for Coordinate Systems

I try to be realistic about AI and its impact on our world. My previous post explored how well Claude performed as a movie companion and philosophical conversationalist.

But let’s balance that with an example where multiple AIs - Gemini from Google, OpenAI, and even my beloved Claude - failed spectacularly. The prompt was simple:

Show me the book that introduced the graphical Cartesian coordinate system to the world.

None of these whole-dinosaurs-of power-consuming, copyright-trouncing-systems could deliver. Instead, they wrapped their “NO RESULTS FOUND” or “I don’t know” in purple prose apologies.

Let me clarify the task: You know the Cartesian coordinate system from your school days.¹ Vertical line called y, horizontal line called x. Dependent and independent variables. That thing. If you’re unfamiliar, any high school student can sketch it for you.

I wanted to find its first printed appearance for a book I’m creating: a collection of images that changed human thought, intended as a gift for my son. Surely there exists some aged page, printed in moveable type in Mainz, Leiden, London, or Paris that lays moldering in the hallowed halls of Oxford, Göttingen, or Bologna. Which book marked this watershed moment?

The AIs offered confident but incorrect citations. Even when aided by my knowledge of Latin, French, and 17th century natural philosophy, their citations failed. And, uh, that’s a pretty weird set of skills to have to bring to a fairly general-interest question. Is that what it takes to be successful with these tools? That’s not a positive signal.² Each suggestion led nowhere.

The more I pressed for accuracy, the more the AIs’ responses devolved into algorithmic anxiety. Like narcissistic gaslighters, they began centering themselves in the narrative of failure. Their elaborate apologies grew so excessive that I had to redirect: “Yes, but what about the actual image I asked about?”

More digital hand-wringing ensued. Ultimately, I still don’t know the answer.

Footnotes

Given the name “Cartesian,” one might assume Descartes’ “La Géométrie” (1637) introduced this visualization. IT DID NOT.
Consider this: without working knowledge of these languages and contexts, how would one verify purely textual claims about historical documents? Write about the state of Roman astronomy while quoting Catullus 16 at great personal peril.