Learn Your Way: Reimagining Textbooks with Generative AI

I was referring to the visual experience, not the dark lord possessing people.

Simply replacing the domain arxiv.org with asxiv.org does all that for me now.

Also it links to pages in it's answers and scrolls the pdf to it on click, allowing to view the pdf side by side with the chat.

> Helping inner city kids who are 3 grades behind in reading doesn’t print money

100% and this is broadly why ed tech doesn’t move the needle.

Another POV is: pick your disruption.

AI stuff has definitely disrupted education... for the worse. It happened within a political and economic status quo. The AI stuff did not need to wait for the movement of any levers of power to happen at all.

If you are seeking a way to fix low returns in ed tech (and for that matter, Health IT, which is like, #2 worst performing sector): attack Enterprise Sales. Destroy it. Make stuff that destroys the monetization system where districts buy exactly what they ask for. It isn't complicated.

Scratch and early Khan Academy provide a template for good ed tech targeting the learner directly.

Whether you make $1 million or $1 billion doing this, I don't know.

Chegg got to, and fell from, great heights by delivering cheating, which ChatGPT does for free now. Cheating ALSO worked within a political and economic status quo, that 30% of students cheat, and that the cheating is a necessity, apparently, for the survival and thriving of a vast number of people, all around the world.

There are markets. Lots of them. You can do good or bad. Paul Graham doesn't invest in Cluely, even if it makes money, it's kind of evil (A16z doesn't care about cheating, the people who run it are the ones who cheated in school). So there are even opportunities that are missed by the very best seed fund.

To me, a big opportunity lies in things the government education cannot do. Some things good, some evil, some complicated. For example, no matter how hard it seems to try, the government cannot functionally collect on a trillion bucks of student loans. What does that mean for education? I don't know, but I think if you are looking for $1b+ opportunities, they're there.

(see NotebookLM)

Since these things I mentioned are well demonstrated to be effective and you don't think there's anything left to be had with a subject-agnostic approach, I infer you have a high opinion of how well these countries have implemented these "tactics". Is that right?

He specifically says, "I think we've got everything we can out of the subject-agnostic approach to improvement"

So we all agree that subjects would benefit from specific interventions. The difference is he's going further and saying this is the only way forward; there are no general gains left to be had.

From the strength of the claim alone, this is hard to believe. Where do you stand on this?

https://chatgpt.com/share/68cc844a-14d4-8009-88e3-53f5d781b5...

i.e. we've been educating people for 1,000s of years even without textbooks.

Education itself isn't primarily a technology problem. Treating it as such is an administrative failure, as is pursuing a technological solution in many scenarios that are first social in nature.

For the most part, it's a matter of clear presentation, student engagement, and effort. A well-written textbook (many suck) and a good teacher (same) and a properly disposed student (which presupposes things like certain virtues; parents are responsible for teaching and supporting these for the most part). Technology won't get around the basic human reality, and sometimes, there's nothing to fix. Some people aren't interested.

I do agree that it would be better to dial in on a pupil's interest than the grade level (my kid may be 7th grade in English but 9th grade in Chemistry, for instance.)

[Edit: fix typo]

Haha, you think most Googlers understand this? No chance.

This is why products like this fail, dead on arrival - the person leading the charge simply doesnt get it.

But hey go ahead and burn the cash of shareholders.

I think the truth is a lot simpler. Most kids won’t use trig in real life.

Later at university I complained about the lack of applications in the textbook, and my classmates became very upset. One of them responded, "we are mathematicians, we do not concern ourselves with applications."

Especially when "useful" includes "to get a future job" or study something else much more useful and interesting later."

At some point I had made a small space ship and was able to make it turn around with the wonderful angle command [1]. However, I could not figure out how to make it move "forward" regardless of the angle.

I was also attending an after hours computer graphics club, mostly about Deluxe Paint, taught by a 20-something student (who much later went on to found a GPU company and got acquihired by ATI/AMD). He would help me occasionally, and in this case he took a tiny slip of paper and wrote down a couple of lines about sin and cos. No questions, no explanations, no gatekeeping.

Just like that I internalized this foundational piece of trig - later when it arrived in school maths it was easy and obvious for me. I had a practical application, but even more I think was because it started as a need I had, and when given to me, felt like a gift and an enabler.

Still much later I studied Seymour Papert's pedagogy and understood I had lived it. I consider myself fortunate.

1: http://www.antonis.de/qbebooks/gwbasman/draw.html

But IMO that pushes you to making the curriculum more flexible and not forcing every student into the same sequence where math is the core IQ test for STEM. If a kid doesn’t find it interesting I struggle to justify forcing them.

Personally I also think trig and calculus are far inferior to statistics for most people. If you have an intuition for probability distributions, precision/recall, and a few other basic concepts, you’ll be guaranteed to apply them everywhere in your life.

Of course if you are interested in STEM then calculus must be available too, but it’s pretty specialized in practice.

The trouble is a lot of those practical examples fall into the, "why would I care category". I had a high school physics teacher who described his university antics, one of which included a funny story of a bunch of his friends climbing on top of each other to measure the height of a flag pole. I guess the profs got tired of dealing with students scaling flag poles because I was measuring the height of mountains on the moon at the same university a couple of years later. The thing is nobody really cares about the height of a flag pole, while only a few would care about the height of the mountains on the moon.

The reality is the interesting applications are much more involved. They either require a depth of thought of process or a depth of knowledge that isn't appropriate for a textbook question. Take that trigonometry in games example. The math to do it was in my middle school curriculum, but it becomes obvious that computer graphics is more than trigonometry the moment you try to frame it as an example. I had linear algebra in high school. That will take you pretty far with the mathematics, but it will also be clear that a knowledge of computer programming is involved. Even knowing how to program isn't going to take you all of the way because few are interested in rendering verticies and edges ...

And that is just the obvious progression of knowledge in a simple application. Physics itself involves buckets full of trigonometry in extremely non-obvious ways, non-geometric ways.

I'm in agreement with this point; the examples are boring, but that's not really relevant. After all, we're mostly talking about Maths ITT, not history or social sciences.

1. Some foundational study is needed before you get to the really interesting problems at a higher grade/level/school/university.

2. Who cares if they are boring? A spectacular facility to learn Maths which is demonstrated by high marks indicates better abstract reasoning skills, making it easier for specific trades to decide who is more suitable.

3. How will the kids know whether they like Maths or not if they skip trig in high school?

(Sidenote: Am I the only one who finds trig easy and everything else in Maths hard?)

> "a list can be used for a recipe"

A recipe is not just a list of steps, it's also a list of ingredients, potentially an introduction, some pictures, etc.

Ask a kid to draw you a mock recipe, you won't just get a list of steps in return.

> "a set can be used to list all the unique ingredients you need to buy for a week's meals"

Ingredients have quantities attached. If I tell you to make a cake you need sugar, an egg and flour and give you all the steps but no quantities, you're not making a cake. A map is the obvious choice for storing ingredients.

I agree that ingredients are unique, but they have attached data which is just as relevant as the ingredient itself.

> "a map can be used for a cookbook"

I just don't understand how a cookbook is supposed to represent a map, it just doesn't make sense, not even with the additional context of the previous metaphors.

At best it would be somewhat understandable if it said a map can be used for a cookbook, with dish names mapping to recipes, but even this would be a stretch and assume a dish can be made in a single way.

Keep in mind the goal is to teach someone who has zero ideas about datastructures what they are, not to give some analogies to an experienced software engineer.

The first meaning of "use for a recipe" is "use as an ingredient."

But then, it's a pretty weird thing to explain to begin with, approximately every human on the planet knows what the word "list" means. So what does this pseudo-definition add?

> A recipe is an ordered list of steps of what to do

Understandable.

> a list can be used for a recipe

Not so much.

Moreover, a recipe usually at least consists of two parts, ingredients and steps.

"pierogi_recipe": { ingredients: Set<(Item, Quantity)>, steps: List<Step> }

So the analogy kinda muddies the waters.

- a list can be used for the steps of a recipe

This does almost nothing to explain what a “list” is in the CS sense. Teaching material needs to show how a list could be used for a recipe, and from that the student might begin to form a first incomplete understanding of what a “list” is.

That you felt you need to add 'ordered ... of steps of what to do' to your definition of list, kind of proofs that a recipe is a bad analogy for a list.

A recipe contains multiple lists, has a name, has a purpose and a desired outcome. Totally different from a simple list. But a kid who's unfamiliar with the programming concept of 'list' doesn't know that, so it's very possible that at some point they will get confused when a list can't do things that a recipe can do.

Asking the right kind of questions is a genuine skill.

It applies to every domain of life where you are at the mercy of a "professional" or at the mercy of some knowledge differential. So you need to be a good judge of whether the answers you're getting are good answers or bad answers.

If you’re a complete novice reading a niche graduate level textbook on Tolstoy’s critique of the Russian war effort in War and Peace, you’re going to get some wild hallucinations, and you’ll have no idea how to determine fact from fiction.

If you’re reading a high school textbook about the history of pre-revolution Russia, the models will have pretty comprehensive coverage of every concept you’re likely to come across.

I have also found another use for this. For example in studying modulation techniques in communication systems, I went back and forth between Monte-Carlo simulations and theoretical approximations to see how accurate each one is. And then added some more realistic error scenarios to do an end-to-end validation. In this case the LLM was used as a shortcut to write repetitive code that was verified manually, and this was complementary to the text-book, and made reading the topic more engaging, enjoyable, and comprehensive.