AI, simplify
I spoke with Michael Carter a few weeks ago, an instructor at St Lawrence College, in Kingston Ontario. Michael is enthralled with AI, and it’s potential in the classroom. He’s getting his students to use it in many different and exciting ways.
In our conversation, Michael made an observation which might be the simplest and most profound statement I’ve ever heard about teaching:
“The biggest barrier to learning is, ‘I don’t follow’.”
This statement cuts in so many ways, but the fundamental insight is that a learner stops learning once they don’t, won’t, or can’t follow. And once that happens, any more time and energy they spend in the classroom, finishing a section of a chapter, is (probably?) wasted. It’s certainly not optimal time spent.
(By the way I include myself as a learner, I expect you are too).
I don’t follow.
I’m not an educator. I am, among other things, a technologist thinking about ways in which technology can help humans learn and teach.
But, I’m also a father, and have spent enough Sunday evening hours at the dinner table, frustrated child beside me, thinking about how to translate a homework problem or concept into language that will land. Language they will follow.
Why don’t we follow?
A learner might stop following for myriad reasons. They understood what you said before, but they didn’t the understand several words in the next sentence; perhaps you spoke too fast, or the text was too dense and too long; maybe the video was presented in a way that bored the student silly. They are auditory, not visual learners. Vice versa. They read too slowly, and have become frustrated; or too fast, and have become distracted. Maybe the material wasn’t connected to anything that matters to them. Maybe they were hungry, or deeply stressed. Perhaps your use of semi-colons has annoyed them.
Whatever the cause, the learning stopped at the moment of “I don’t follow.”
A series of moments
The corollary is that education is largely about crafting a series of moments where a learner follows, from one thing to the next, gaining a skill, mastering a bit of knowledge, understanding the next step along the way. (Though some would say the most successful educational experiences happen when the learner and instructor are taking turns following and leading.)
“Simplify, please”
Off the shelf AI is already spectacularly well-suited to solving parts of this “I don’t follow” problem. We don’t need virtual AI teachers or complex personalized learning. ChatGPT is already very good at repackaging information into different presentation formats. It can simplify and summarize, it can reframe your math problems to focus on basketball, and reframe questions, and make suggestions.
There are many more exciting and robust ways that AI can and will be used to help teachers and learners, but right now one of my favourite uses of AI is just a very simple prompt:
“Simplify, please.”
Here’s an example of simplifying a paragraph from a free digital textbook, Allied Health Microbiology , originally published by OpenStax, revised and updated on Pressbooks by Linda Bruslind, at Oregon State University:
Quorum sensing involves cell-to-cell communication, using small diffusible substances known as autoinducers. An autoinducer is produced by a cell, diffusing across the plasma membrane to be released into the environment. As the cell population increases in the environment the concentration of autoinducer increases as well, causing the molecule to bind to specific cellular receptors once a threshold concentration has been reached. The autoinducer then diffuses into the cell, often binding to a specific transcription factor. This produces a conformational change that allows the transcription factor to bind to the cell’s DNA, triggering expression of specific genes.
That’s a pretty dense paragraph. I’ve have created a little GPT bot that will simplify that paragraph for me. (Try out the bot, let me know what you think — you’ll need a GPTplus account, sorry). Here’s what that chatbot did:
I’m a good reader, but don’t know much about microbiology. My eyes glaze at the top paragraph; but if you ask me what Quorom sensing is now, I could tell you. “Quorum sensing is cell communication through molecules.” There’s more detail I need to understand, but:
I follow.
I’m ready for more.
Simplifying is trivial for AI to do, and AI is very good at it.
Simplifying is also a trivial use of AI: but that’s what makes it most exciting. With very little lift AI can help unstick those sticky “I don’t follow” moments. My bot is just a little example. There are so many more ways we can think about AI in the context of helping students say, “Yes, I follow, what’s next?”
Struggling learners and advanced learners alike
What’s interesting here is that you can imagine this helping students at both ends of the learning spectrum. You can easily imagine this helping a learner who says “I don’t follow” because they are getting lost in the language of the original paragraph.
Just as easily, you can imagine the precocious student bored with grade 4 (or 1st year) science class, keen for more. Perhaps they’re not quite ready for college-level books. With a little application like this, suddenly inscrutably advanced content becomes accessible to the curious mind.
Learning is about struggling to understand the moment before, and making the leap to the next moment. AI can, and should, help.
Are you interested in AI and education? Please get in touch.
Here are three articles that caught my eye this week:
Cyborg: Using Co-Intelligence: How I used AI in my book about AI (Ethan Mollick)
The last millimetre problem: Apple is Ono-Sendai, and we’re pushing beyond real-time to anticipatory interfaces (Alistair Croll)
Introducing Devin, the first AI software engineer (Scott Wu)