Teaching and Learning Machines

We can either do the learning and heavy lifting or have the learning done to us.

ELT Pros Linkedin | Videos | Blog | Printables | ELT News | TpTs | YouTube

I return again (see one of my previous posts) to the subject of ChatGPT and LLM - large language models. It’s all the buzz and I’m glad for the discussions people are having surrounding the use of ChatGPT in education. It can’t but help though I’m a bit pessimistic, believing that so many won’t put the brakes on and will be pulled down the ChatGPT rabbit hole. Let me explain and share a few thoughts.

I spent about 4 hours today, using one of the many new tools harnessing AI - Rytr. I spend a good amount of time using educational technology - call me an early adopter. I’m just damn interested.

Rytr truly is impressive. It takes ChatGPT to another level, making it so easy to generate text and then rewrite that text to avoid plagarism, insert your own preferences and references into the text and organize it as you wish - all with a few clicks of the button. Voila! Magic.

Reminds me of how I loved those old Texas Instrument calculators with algebraic functions! Whoooooosh and poof! A magic genie giving you the answers. Except here, you don’t just get answers but you get the whole damn prelude, the meat and potatoes, not just the sauce.

But what struck me most about Rytr and too ChatGPT, even its playground, is its banality. It is dry. It produces souless sap. Yes, many think it can make cute poems, songs, write in the vein of Hemmingway … but I’m not impressed. Shakespeare it ain’t yet. If you are a reader - you’ll note after the first wave of amazement has passed, how dry, how bland the text generated, truly is.

And that’s how it is and might always be with AI that tries to pass itself off as human generated. Maybe when Open AI releases its next iteration, said to be a model trained on trillions of parameters (tokens) rather than at present only 175 some odd billion. But I’m not convinced, more will equal better.

Skinner tried with his teaching machines. (See Audrey Watters brillant book and essay on this topic). But basically his machines and the many subsequent forays of ed-tech into the realm of teaching have been hard core failures.

Why? Because they are based on models of education that value information, data over its application. At present, education is way too far into that model. Valuing cut and paste sound bites over whole heeled creativity. We teach for a test, a set of expected responses and answers. We’ve standardized everything to death. Our schools, medical school to beauty school - they want you to fill in the blanks. But life ain’t like that, I’m telling you. So I reject wholly this kind of education. It’s why we have so many idiots in seats of power - they haven’t been taught to think for themselves and to questions, test, evaluate, apply their learning in creative ways …

“Where is the wisdom we have lost in knowledge?
Where is the knowledge we have lost in information?”

T.S. Eliot,The Rock

Now, how should education react to ChatGPT? That seems to be the 6 million dollar question at the moment. Should it batten down the hatches and prevent students from using ChatGPT? Like we do and did with cellphones and access to Google search? Or should education reinvent itself and allow students to use ChatGPT and rewrite the assessment rulebook?

I’m for the later. So is the CEO of OpenAI. You won’t be able to defeat this kind of tech. But what you can do is realize its limitations, that it generates soulless garble. And so, let’s expect more than soulless garble from our students.

Today, I listened to Chomsky respond to the question. He basically thinks ChatGPT is a big, fancy, hi tech plagarizing machine. Skinner’s teaching machine created by Dr. No. And he’s right to a certain degree. It plagarizes pretty good.

But let’s ask an insightful question - why do students want to use ChatGPT to get the answers? Perhaps it is because we’ve failed. We’ve failed our youth. They see education as a game, a credential, a hurdle, a problem to bypass. They don’t see value in learning for learning’s sake. They don’t see that they are selling their soul to the devil when they reject learning something themselves, instead of asking a machine for a result to submit as their own. I mean, it is the same with steroids and sports. Shortcuts. The outcome is the only thing important. Appearances. Not the real thing itself - YOU, what you are made of, inside.

I said I was pessimistic, I am. Our lives teach us who we are. Our teaching will teach us who we teachers are. Teachers will opt for quick things, efficiency - instead of their human hand and heart. I’m pessimistic in the sense that most teachers won’t stop and ask the right questions regarding their use of ChatGPT. They’ll get sucked into its vortex and technological “ease” and into the inhuman realm of banal content. That would be teaching’s loss.

ChatGPT as a personal productivity tool doesn't interest me. It might be a "game changer", but I'm not playing a game: I'm engaged in deeply human work, and I need to bring the same level of authenticity to it that I ask students to bring.

So, we can see that ChatGPT allows people to pretend to be something they aren’t. And that isn’t good. It can never be. But just like God’s dilemma with the Garden of Eden, we shouldn’t fence it off from students. Freedom is the real lesson here - freedom and its twin, responsibility. Maybe ChatGPT is a good lesson for us all - that we alone and not machines, must be responsible for our learning. For we alone are the ones that will bear the consequences and disasterous results if we reject the responsible use of technology.

And to end on a lighter note and piece of social commentary - but also relates to the ChatGPT generation forthcoming …

All about teaching, teaching English, ed-tech and learning language.
David Deubelbeiss