Here’s my current approach to explaining AI and language models:

When people hear “AI,” they can get stuck on the big, abstract (and important!) ideas around “intelligence” and “artificial.” Plus there’s usually a bit of anxiety thanks to all the dramatic storytelling about AI since the 60s. Thus, I’ve been avoiding the word “AI” in classes for a while, but using “LLM” hasn’t really worked either.

Recently I’ve begun introducing the concept as “literate technology,” or LT. Reframing AI as LT focuses on what the system actually does well: reading, writing, and making sense of language.

We don’t need “user interfaces” or write code to use LT. You can ask questions, describe problems, or draft reports the same way we did before computers: by reading, writing, and talking. Our technology is now literate, and that’s a big deal.

This approach has been effective. Framing AI and LLMs as literate technology makes them feel less like mysterious black boxes and more like practical tools with clear boundaries. In this light, literate technology becomes a way to make literacy itself a computable resource and the value of literate computing is (usually) perfectly clear.