Categories
digital rhetoric

ChatGPT and the model of understanding

Writing a post analyzing the moral panic surrounding the latest digital media technology is nearly as boring as participating in the moral panic itself. Let’s not worry too much about ChatGPT. In a decade it will probably seem as quaint as Clippy. So I’ve got a tangent here and this Harry Potter-esque post title.

I’ll start with the model of understanding. This is a term used in AI research to describe how machine learning processes (i.e., algorithms) operate to develop models that allow AIs to identify objects, generate media, and other tasks. Basically if you want an AI to identify bicycles, it’s going to need a model of bicycles. Generative AIs also need models, so that when you ask DALL-E 2 to produce an image of a pig ballerina in space it knows when it has generated a string of binary digits that will appear as something that looks like the requested image (provided you’re using the right software, of course).

These generative AIs are emulators. They have an understanding of emulation but nothing else. You might say that they’re faking it. They don’t understand the significance of the media. They can make images and texts with pigs as their subjects. They can provide a definition of a pig. But they don’t know why a pig is significant, even though they can produce a text explaining why pigs are significant. If you ask ChatGPT to write an essay on Hamlet, it can do that, but it doesn’t understand the play or have an interpretation of it.

For example, when asked to “explain deleuze rhizome” (or actually “explain delouse rhizome” because autocorrect), ChatGPT produced this.

In their book “A Thousand Plateaus,” Deleuze and Guattari introduce the concept of the “rhizome” as an alternative model for understanding how things are connected and related. They describe the rhizome as a type of underground stem or root system that is constantly branching out and connecting to other things in a non-hierarchical way. Unlike a tree, which has a single trunk and branches out into a hierarchical structure, the rhizome is a network of connections that can be entered at any point and which has no central authority or hierarchy.

The concept of the rhizome has been influential in fields such as philosophy, literature, and cultural studies, and has been used to understand a wide range of phenomena, including networks, systems, and social and cultural organizations. It has also been used as a metaphor for the way that knowledge and information circulate and grow in contemporary society.

ChatGPT

To be clear, this isn’t just cribbed from Wikipedia or something. Of course a professor would be happy with a graduate student explaining the rhizome so succinctly. Hell, I’d be happy to offer this text as an explanation for students (as far as it goes).

That said, the obvious difference between a student struggling to write about the rhizome and ChatGPT’s breezy explanation is that the former is trying to articulate an understanding of a concept and the latter is simulating an understanding. The former may have a poor understanding of rhizomes but the latter has no understanding of rhizomes at all. It is simulating understanding. Or, to quote from another essay I asked ChatGPT to write, we might argue that we are in a hyperreal situation.

Baudrillard argues that modern society has reached the fourth stage, in which the distinction between reality and simulation is blurred and the hyperreal has become dominant.

Fair enough, but that’s why we need a new theory, a new way of seeing, where the distinction comes into focus. The distinction exists here in the material process of composing, of course. That’s where we need to be looking.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.