Yeah, frankly I think education is one field that can stand to benefit the most from AI. I mean every single child or student or learner in the world can have a super high quality teacher that is working with them throughout life. I mean that is a game changer. When we launched edX in 2011, I still remember in an opening speech at the announcement conference with Harvard and MIT, you know, I said look, I'd like to reach 8 billion people around the world.
So I know this is an interview but I'll give you my opinion. I always felt the open source part of Open edX was just one part of the story and I've always been frustrated with the fact that the courses themselves are not shared between platforms. So we have this whole community of people launching Open edX platforms and we don't really share our courses.
You know, I'm into sharing in a big way. My initial course on edX out of MIT we divided that into three courses and I've open sourced the three courses under a CC BY-SA open source license.
So it's a shared course in that sense. But that said, I think your question had to do with everybody creating courses on Open edX instances. Isn't there a way to somehow syndicate or somehow share all the courses? And frankly I would love to see every single course that has been created, maybe they all single sign on into edX.org or something, and anybody can go and learn and take any course from anywhere. And we can find ways, you know, I know people are spending money and livelihoods have to be supported and so on, so maybe if people pay for a course there's a way to revenue share across all of that.
It's not just that, I would really love to be able to pick a course from MIT and like take a unit and use it in my real life programming class with high schoolers, and maybe modify some content and share that content again.
Maybe MIT will pull some content. You're speaking very much like a software developer.
And so in software we are really well versed in borrowing pieces of content. We use libraries, whether it's various Python libraries or AI libraries or IO libraries, you know, the culture of sharing, we share libraries, we share code. Very rarely do I write, if I write a new program, very rarely do I start writing from scratch. My style is I usually pick various bits and pieces and sort of club them together and build a program and then iterate upon it. That same thing hasn't happened in content production. Content still is a very bottoms up activity and there have been many efforts at writing textbooks through an open source model and very few examples have worked, very few examples.
I've written a textbook and my colleague and I wrote a 12 or 1300 page book where we borrow and reference other pieces of work, but largely our work. There's a real reluctance and a real anxiety for people to share bodies of their work where somebody creates another course based on other pieces of work. Testing is hard. I think that's the other issue. With code, you have unit tests and so you can test whether something that you've introduced still works with a testing library or with regression tests.
With content, it's very hard to have regression tests. And often with content, in creating a book or a course, there is a viewpoint and to my knowledge we don't have tests that will test whether a new piece of content that's embedded violates that central thesis or the philosophy that I had through the course. Maybe AI will fix that. But until we get such a test, I don't think the sharing of content and creating new works based on previous pieces of content will ever happen.
But isn't it happening right now with AI? I mean, this library of content that you're describing, the big AI companies have already illegally swallowed all of this content and are redistributing it in the form of Gen AI.
Yes, they are doing that and a lot of the people that produce the original content are extraordinarily unhappy. So the New York Times for example sued OpenAI. And I think until all of that can be figured out, I don't think creating content in a collaborative manner will work. We haven't figured out how revenue gets distributed when you create shared content. We haven't figured out how credit gets distributed. We haven't figured out how to create a textbook or a piece of art that has a unified theme based on sharing pieces of content. So I think there are a lot of things to be figured out before something like that can become truly prevalent. I think the closest open source shared piece of work that I've seen is Wikipedia, where with Wikipedia it's truly a big open source effort, but there you are creating a gigantic encyclopedia, which by its very nature is easy to create through sharing. But a unified essay, I think it's very hard to create a unified essay extolling the virtues of democracy for example, it's much harder to do with 100 people collaborating and providing three sentences each.
Yeah, I'm constantly astounded by the stability of Wikipedia. I remember editing Wikipedia 20 years ago, 2005. And it seems to me that Wikipedia hasn't fundamentally changed since then. Whereas companies have come and hyped and bankrupt since then. I mean, Facebook was created like a few years later, I think it was 2006 or 2007. Yeah, I don't know anybody who uses Facebook anymore, especially among the younger ones.
I think the older generations still seem to. So yeah, it's way past hype I would say, though obviously there are other products by Meta. But yeah, so they must have gotten something right there.
