LLM-Integrated Multivariable Calculus Course(calculus.academa.ai) |
LLM-Integrated Multivariable Calculus Course(calculus.academa.ai) |
As someone who has self-taught most of my skills both before and after AI, some deep feedback: I don't want a fixed piece of content when learning with AI, like a video or blogpost or book—unless I'm completely new to the subject, and even then maybe not.
The reason is that some parts of the topic will be naturally easier or harder for me. When I use AI I tell it everything I know and understand and start working from my most burning questions and misunderstandings. This lets me cover the maximum amount of non-redundant ground in regards to my understanding.
We have this amazing new technology and you're conforming it to models of schooling (like the Prussian model) which are one or more centuries old. The technology is so powerful that it should allow you to completely reshape education, not merely replicate the status quo.
Definition: the AI tutor is some AI that can generate these videos.
If that generation happens in real time, it's a complete reimagination of education. It's literally the AI private tutor.
But before you can do that in real time, ask yourself: let alone real time, can you even generate a video, let's say, in 30 minutes? That's what we explored, and we found some great things. But it's not real time yet. The only practical way to publish what we'd built, though, was as a static video paired with chat.
The complete reimagination of education is on the way.
Most literature seem to indicate that whatever the manner of use, "friction" while learning ultimately still helps. My first "experimental" touchpoint is using LLMs as a called "socratic tutor".
Edit: because fat fingered submit while typing this on a split keyboard
That narrator drives me nuts though. There too many “like” “you know” “so” “down here”. Comes off grating, I don’t think I could listen to multiple hours of this.
At the same time, thousands of low-quality resources overwhelm good content.
Integrating with an existing resources like MOOCulus could add a ton of value, in contrast, but wouldn't have the promise of making the creators money.
Being able to create manim animations at scale is a value-add, but doesn't seem enough of a value-add to create a business which does anything other than active harm. But it seems to be trying to be one.
"The videos are LLM slop." - No, they're not. Watch them.
"It's dogshit that no one can understand." - That's not true. Try it.
- The voice is clearly TTS, which I think really loses something. Not having variation or stressing particular parts with intonation is a big deal in teaching. The beginning of lecture 5 has a pause in a weird spot (surface).
- The intro to the first lecture is already unnatural: "And that's of course now we're going to start working with functions of more than one variable."
- The scaffolding in lecture 1 is off to a bad start. The lecture tries to rope the student in with a question, but the question uses the notion and notation of a vector that has not been introduced yet. This reeks of a prompt "use a (motivating) question to introduce the topic", but a question that cannot be understood by a student does not help.
I'd provide feedback on how to make this better, but to be frank, I don't want to see this made better. I'd like open education made better. I'm on the other side here. Even if I did, I'm pretty sure you wouldn't listen.
In my career, I've done both open and closed models (more of the former in recently, more of the latter early career). I'm open to both, but education should be open, for a whole slew of reasons I can enumerate.
Our videos are not some MP4 files sitting on a disk. We see our content like software. We review it, version-control it, and maintain it.
We're two PhD students and close friends since our undergraduate years.
I'm Sina: https://sinaatalay.com
And my co-founder, Apo: https://geduk.io
We created the technology so that the content can be updated and improved over time. If we make mistakes, we'll fix them and continue maintaining the material. These aren't static MP4 files recorded once with a camcorder.
I highly encourage you to take "didicatically useful" as a bar to aim for, rather than "not making mistakes".
I think it's a cool tech demo, but if you're positioning this as a course that's ready to be consumed by students, I think doing the quality control is not optional.
With manim, it's probably legal (I think MIT license?), but in general, it makes the world a worse place.
Speaking of which, it happened before, too, in the 1970's with personal computer software.
That's very clear. That's why we have spyware, ad tech, WWII-era German chemical weapons, the East India Company, Oracle's business model, fintech, ...
Ask anyone 40 years older. Regrets focus on time with family and meaningful work. Few people think: "Did I innovate enough?"
> We simply can't open source this because we're rewriting and rebuilding it every day.
I don't quite see the logical connection in this sentence.
Everything I've worked on in the past 15 years is open-source. I suspect you'd:
- Be able to find modest amounts of funding to do this in the open space (e.g. within the academy, as a teaching professor, adjunct, or similar) -- enough to sustain yourself, but without a promise of riches
- Be able to "innovate," make more E($), and do less harm in corporate America
It would also make it easier for competitors to build products that compete directly with us, such as https://studio.academa.ai. That would make it harder for my startup to succeed.
I'm not building this for the sake of making money. I simply need enough protected space to innovate competitively and give the company the best chance to succeed. If it does, I'll be able to keep building and pushing education forward for years to come.
Personally, I just used chatgpt the other day to _finally_ understand why a dot product of vectors is equivalent to cosine similarity. Mathacademy's lesson is very good but the explanation wasn't enlightening for me, it took chatgpt breaking it down into tiny bite-sized chunks, patiently re-explaining what a cosine actually is and where it disappears to in the dot product, making little animations of vectors with sliders that I could move myself to try stuff with differently shaped and rotated triangles, for me to _actually and viscerally understand it_ and after spending probably 2 hours on this simple concept, now I _actually_ feel like I understand it rather than having just memorized it. So I feel much more confident that I can build deeper understanding on top of that.
Plenty of open source is updated every day.
You had no problem taking advantage of Manim to profit, but when it comes to returning the great favor done to you by Manim being open source, you don't care?
Okay.
You may not believe me, and you may think I have ulterior motives, but I don't. My only goal is to build technology that meaningfully advances education.
I believe you entirely. That comes across in everything you say. Huge harm has been done in education precisely by people with similarly amoral motivation. History of education is filled with case after case of amoral motivations leading to immoral behavior.
Other industries have this more extreme. Fritz Haber is a biography worth reading.
The comments you're making aren't very honest -- I suspect with yourself more than with anyone else. For example:
> Maintaining an open-source project requires significant time and effort.
No. It doesn't. You need to flip the "public" bit on github and post a "caveat emptor" shingle. The effort you're cloning -- manim -- did exactly that. The result was a synergistic, friendly, manim-ce.
And the author has been able to work full-time in this space, his entire career, without doing anything evil, unethical, harmful, or unpleasant, and strictly advancing the world.
You're competing -- rather than cooperating -- with folks like him.
> I simply need enough protected space to innovate competitively and give the company the best chance to succeed.
You don't seem to understand how this works. A few points:
- Education is a highly-regulated industry for good reasons (student harm, societal harm, privacy risks, etc.), and education research goes through IRBs for similar reasons.
- Startups can work on IP, but usually -- and especially in education -- networks, connections, support, etc. are more important. It's hard to build an IP moat.
- Startups have a 95% failure rate. I'd give this one more than that.
A few corollaries:
a) With the current model, you'll have a hard time building partnerships with the people you need, and indeed, people (myself included!) will be rooting for you to fail. That weakens, not strengthens, your odds of success.
b) An example of another way to accomplish your stated goal is a five-year research grant, or a position which supports doing this kind of work (e.g. instructional staff at your university). Those have >> 5% success rates.
c) Open business models aren't less likely to be successful here than closed ones. Red Hat sold for $34B.
Otherwise, my best wish for you is to fail fast, and to learn a lot from the process.