In a dystopian future, a book — a primer — educates Nell, a young orphan. Nell does not go to school; the primer becomes both her parent and her teacher. She talks with it, and the primer answers. It educates Nell. The story is Neal Stephenson's 1995 novel The Diamond Age: Or, A Young Lady's Illustrated Primer. In the novel, the primer is a marvel of nanotechnology and artificial intelligence. Today, we might know it as a Large Language Model, or an LLM.

LLM Primers

I was reminded of Stephenson's primer while talking with colleagues about LLMs, such as ChatGPT, tutoring students. Today, we can ask LLMs for more information about, or for an easier-to-understand explanation of, almost any topic. As with the primer that Nell uses, we can converse with LLMs to get answers and learn more about topics of interest. I use that verb, converse purposefully.

Recently, I asked ChatGPT to help me find a book that I remembered scenes from but could not remember the name of. It felt like a real conversation because ChatGPT replied with relevant follow-up questions and suggestions. It even wondered if I could be conflating parts of two books and gave me the names and summaries of both. Actually, it was correct — I was conflating two books I had read. That is pretty incredible, but more to the current point, because of this back-and-forth, my interaction with ChatGPT was like having a conversation with a bibliophile friend. It really seemed like a conversation with a human. LLMs can have conversations like this on almost any topic.

Imagine students interacting with an LLM primer for most of their formal K12 education. Students who are ready easily jump into more advanced content. Struggling students get help anytime and anywhere, at any level, with no waiting and no judgment. Like Schwarzenegger's terminator in the movie T2, LLM tutors would never leave, never shout, never be too busy. Every student could have a conversation about any part of any content, digging in as far as they wanted, or getting as much help as they needed to acquire all the basics.

For those who are interested, the full Sarah Connor quote from T2 is:

Watching John with the machine, it was suddenly so clear. The Terminator would never stop. It would never leave him. It would never hurt him, never shout at him, or get drunk and hit him, or say it was too busy to spend time with him. It would always be there. And it would die to protect him. Of all the would-be fathers who came and went over the years, this thing, this machine, was the only one that measured up. In an insane world, it was the sanest choice.

Assuming equitable access (a huge assumption, I know), LLMs really could have an enormous influence on students' education. This is especially true for students from less advantageous backgrounds, but it holds for all other students as well.

Sure, some students could use it like a better search engine and just look up answers. However, the impact of technology is not measured by the least it can do, but by the most it can. Students will interact with their LLM primers with academic and non-academic questions. The LLM primer will remember all of its interactions with the student. The student can personalize it (answer in Pirate on September 19). For some students, it may almost be a friend. We will not see for years exactly how students' use of LLMs will impact their own growth and education, but I think the impact will be profound. For many students, LLMs will be Stephenson's primer.

LLMs are already doing incredible things in education. Sites are offering LLM-powered tools that do almost everything educators need. LLMs are creating lesson plans, assessments, writing samples, presentation slides, letters home, translations of all of those, and differentiated versions of most of those. Some teachers are already asking LLMs to provide feedback on students' writing. In a few years, that will likely be the norm, because it will save teachers a huge amount of time. Using LLMs for all those other things I mentioned above will be the norm not in a few years, but probably in the next year or two.

The output now from the LLMs doing all of the above is already good enough. Technology almost never gets worse; it only gets better, so I expect scope of its impact on education will only increase each year.

How will education change when, thanks to LLMs, the teacher is no longer overworked and overwhelmed? Today many teachers, with the best will in the world, struggle to do basic differentiation for English Learners, struggling learners, and advanced students. What will in look like when they don't have to? When each student can basically do it for themselves with their LLM primer? Or, more likely, the teacher used their own LLM to do some basic differentiation, and the students further differentiate that for themselves as needed?

I have used ChatGPT as a learner. I took a Spanish class at the local community college and used ChatGPT to supplement classroom content. ChatGPT created extra readings for me to practice on. It expanded on grammar topics we had learned in class. When I was confused about something — such as when to use ser or estar — I chatted with it and gained a better understanding of that grammar. ChatGPT significantly aided my learning. And that was an earlier, less-powerful version than is available today.

Not Just Another Fad

Could LLMs be just another fad, with no more long-term consequences than all the other revolutionary technologies that have come before — the VCR, the calculator, or even the internet? I don't think so. Those examples are no longer feared disruptors of the status quo; they are (or were) normal classroom tools. The internet, though, stands apart because it is not a tool that does a single thing. It is a protocol that we do many other things on top of. Calculators replaced rote memorization, slide rules, and long division; the internet replaced VCRs and DVD players (with streaming video), card catalogs (with search), and typewriters and word processors (with online word processors). In fact, browsers and the internet have almost replaced all locally-installed software. LLMs will likely have just as big an impact as the internet. Soon, perhaps in just a few years, LLMs will also be normal classroom tools, just as the internet is today.

Sometimes new tech really is disruptive. LLMs are much more similar to the internet than to the VCR or the calculator in this respect. They don't do one thing; they do almost anything people can do with language. Perhaps the true impact of LLMs should be compared with that of the internet or that of electricity? Okay, I would not go as far as electricity, yet, but certainly the impact will be huge. There is a famous observation that when Apple introduced the iPhone in 2007, no one realized it would create a new type of employment (the gig economy) and destroy the taxi business (Uber, Lyft). People knew the iPhone was revolutionary, but they had no idea just how much it would change the world. Large Language Models will likely have as large an impact, and as with the iPhone, we have no idea today in what ways it will end up impacting society. Education is just one, fairly obvious, area it will have a huge impact.

Caveats

Yes, we have to be careful. Student LLMs must have appropriate, extremely resilient, guardrails. Student data privacy must be a non-negotiable. Accessible content and output must be grade appropriate and aligned with curriculum. In addition to student privacy, we must somehow balance the commercial needs of the companies providing LLM technology with the responsibilities that schools and families have toward their students' education. That will not be an easy job, but just as educators are now benefiting from the power of LLMs, we should welcome students also benefiting from Large Language Models, and help education change to fit this new tool for student learning.

Maybe we should ask an LLM how to do that?

About the author

Chris Spackman has been teaching ESOL in Columbus, Ohio since 2010. Before that, he taught English as a Foreign Language in Japan for 13 years. In addition to teaching ESOL at almost every level, from kindergarten through adults, Chris taught TESOL technology and resources to MA TESOL candidates at Ohio Dominican University. He is currently the ESOL Coordinator for The Graham Family of Schools and an ESL Educator at Columbus State Community College.