Learning a Language part 3 - with video

Well... I ran the HTML class 😂

And something happened that I did not really think about when I wrote the first two posts.

If you read the other posts, the idea was relatively simple.

ChatGPT on my phone would become the live teacher. The classroom on my computer would show me the lesson, images, reading, questions, fill in the blanks, etc. And an ICM structure would sit underneath it all so that lessons and progress don't disappear every time a conversation ends.

We built it.

And I sat down and took the first real lesson.

It worked.

The teacher knew what I was seeing on my computer. I could read something on the screen, answer the teacher by voice, click through an activity, and continue the conversation.

For the first time it actually felt less like "talking to ChatGPT in Thai" and more like I was sitting in a language class with a teacher who happened to also know exactly what was on the screen in front of me.

Except...

The lesson was too easy.

😂

I went through all 13 activities in about 6 minutes. Every one of the activities the computer could actually check, I got right on the first attempt.

At first you could look at that and say:

Great! I passed the lesson.

But that turned out to be the least interesting thing that happened.

Because while the classroom saw that I got everything right, the voice teacher noticed something different.

The lesson was working with the Thai structure ถ้า...ก็...

Basically an "if this... then this..." relationship.

I could use it.

I could make new sentences with it.

At one point I even made a sentence about going surfing if it didn't rain that wasn't one of the sentences the lesson gave me.

But when the teacher asked me to explain what ก็ was actually doing in the sentence, my answer wasn't nearly as clear.

And suddenly we had three different pieces of information.

The computer said I passed everything.

The teacher said I could use something that I couldn't necessarily explain very well.

And I was saying the whole lesson was way too easy.

All three were true. This is where this little language-learning experiment suddenly became much more interesting to me.

Because I realized the feedback loop shouldn't just be: Teacher → Student.

It should be: Teacher → Classroom → Student → Classroom → Teacher.

The classroom shouldn't decide how to teach me. That is still the teacher's job.

But the classroom can watch what happens.

• Did I need three attempts?

• Did I open the English translation?

• Did I fly through something that the teacher thought would be difficult?

• Did I struggle with something that was supposed to be easy?

• Did the lesson say it was testing conversation but actually spend 15 minutes making me click multiple choice questions?

Then at the end, the classroom can hand that information back to the teacher.

And the teacher can hand back something the classroom cannot really know.

Maybe I pronounced something strangely.

Maybe my Thai was understandable but sounded like I had taken an English sentence and replaced every word with Thai.

Maybe I understood something perfectly in conversation but couldn't explain the grammar rule.

Maybe I could read every word on the screen but had absolutely no idea what the paragraph meant. 😂

Those are different things.

So now we have started separating them.

Speaking naturally.
Sentence structure.
Reading.
Reading comprehension.
Vocabulary.
Pronunciation.
Understanding a structure.
Actually using that structure when nobody tells you to use it.

And even something I had considered before: Street language vs classroom language.

This was actually one of the things that started this entire experiment for me.

I don't want to learn a language where every sentence sounds like it came out of a grammar book.

But I also don't want "just talk to people" to be the entire learning method.

There is value in understanding why a language is structured the way it is.

There is also value in knowing that nobody at the food stall is going to speak to you the way the sentence in chapter 4 of your language book is written.

A good teacher should be able to move between those worlds.

So now the classroom itself knows which one we are working on.

Sometimes we might be studying the structure.

Sometimes I might be reading more formal Thai.

Sometimes the teacher just needs to talk to me like someone I would meet outside and see what happens.

And then another thing happened. While we were building lesson 3, the teacher looked at the assessment from lesson 2 and said something like: "I would probably emphasize this part less now because Gabe clearly already understands it."

My classroom stopped.

It basically said: 'Wait. Is the teacher telling me to change the lesson? Or is the teacher just thinking out loud?'

😂😂😂

And THAT opened another rabbit hole.

Because now we had to teach the system the difference between a teacher reflecting and a teacher giving an instruction.

The teacher can say: "That was easier for him than I expected."

That is information.

It can say: "Next time I want to push this further."

That is teaching context.

But neither should secretly rewrite a lesson that my builder and designer are already working on.

If the teacher wants to change the actual lesson, it now has to explicitly say: REVISION REQUEST.

Otherwise the classroom keeps the lesson intact and simply records what the teacher learned.

Which made me realize something else I really like about where this is going.

The student isn't the only one being assessed anymore.

The teacher is too.

Not in the sense of giving ChatGPT a grade.

But if the teacher predicts something will take me 15 minutes and I fly through it in 4, that tells the system something about me...

and something about the teacher's current understanding of me.

If that happens repeatedly, the teacher needs to adjust.

That feels much closer to what a good human teacher actually does.

They don't just keep a scorebook on the student.

They slowly learn the student.

And because all of this sits inside files rather than disappearing into one giant AI conversation, we can keep that history.

Eventually I want every 30 lessons to produce something much more interesting than: You are now Level B1. Congratulations.

I want it to be able to show me:

Here is how your speaking has changed.
Here is where your sentence structure has become more natural.
Here is what you can understand but still struggle to produce.
Here is how your reading has changed.
Here are the things you used to need help with that now appear naturally in conversation.
Here are the kinds of lessons that seem to result in better retention later.
And here are the areas the teacher thinks we should push next.

The funny part is that when I wrote the first post this morning, I was excited because I had figured out how to put ChatGPT voice on one device and a classroom on another.

Then I thought the breakthrough was turning that classroom into HTML and letting my ICM workers build the lessons visually.

Now I think the much more interesting part might be this: 'The classroom, the teacher and the learner are all leaving evidence for each other.'

And every lesson makes the next lesson a little more informed.

My Designer can make an animated visual explanation.

My Content worker can make a story for me to read.

The classroom can see how I interacted with it.

The voice teacher can hear what the classroom can't.

And the ICM can keep the history without pretending that all of those things are the same kind of evidence.

I have absolutely no idea how far I'm going to take this.

This morning I just wanted a better way to learn Thai. 😂

But lesson 3 is being built now.

And strangely enough...the system knows more about how to build it because of what happened in lesson 2.