58,000 Sign Up for Stanford AI Course

Nice try, but 1% of 58,000 is 580.

Did you take the prereqs? :)

And if 0.1% complete the course, that's 58 people. Despite what you may have heard, there are numbers less than 1%. Many people who buy advertisements in publications with, say, 100000 readers have learned this lesson the hard way.

... 580 people ...

Maybe a bit farther than 10 years. I watched CS lectures online back in 2000 for my USC classes.

They can also use the data the crowd is generating, these are big data guys.

Also, students are going to be trying harder because they are going to be competing against the world. Imagine if no in-class student was in the top 1,000?

A regular Stanford student can expect to spend about $50,000 a year [1] or about $200,000 for a four-year degree. In four years, I believe that there will be at least one company that will offer trustworthy certification for a lot less, like maybe $20,000 or $2000. If the primary benefit of a Stanford education is the certification, then in a few years Stanford grads may be like owners of $200,000 cars-- obviously wealthy, but not possessing any significant advantage in transportation over owners of $20,000 cars.

[1] http://answers.yahoo.com/question/index?qid=20090416220215AA...

> you're paying for the grade & the transcript, not the content.

How much is the grading and transcript worth? How much does it cost? I think there's a great opportunity for a startup to just focus on offering transcripts. The hardest part of education is the course content. The grading and transcript part is easy.

I love the concept of open education. But it seems to me that universities who do this are self-destructing. Does Stanford really want to base it's value proposition on its administrative staff? "Anyone can get the knowledge for free. But if you want certification from the Stanford administration staff, that will cost you $150,000. Come to Stanford and get a diploma!"

I just listened to a talk[1] by Ori Brafman in which he mentions that one of the most important factors in professional collaboration and friendship is physical closeness. MIT dorm students were twice as likely to be best friends with their immediate dorm neighbor (6 meters away) than with the dorm neighbors two rooms down (12 meters away). Professional collaborations followed the same pattern. Y Combinator likes to see founders share an apartment. There's probably a lot of value in simply being physically near other people with similar interests.

[1] http://ecorner.stanford.edu/authorMaterialInfo.html?mid=2683

Some variation of this happens in HK, where everybody wants to learn English (because they can't learn as well from their peers, unlike other subject) and English lecture/tutoring places pops up all over the place, and the top teachers teaches, via video conferencing, hundreds of pupils at once. I'm told the top teachers gets paid in the millions.

The appeal here is that the course is free, ergo, _anyone_ with a basic internet connection can take it.

If you require payments, you then have credit card companies and Paypal restricting who (i.e. which countries) gets access to it.

Income from donations isn't guaranteed, and thus provides a weak incentive for universities to provide these kinds of courses. I'd also wager that donations would fizzle out as this sort of thing became more commonplace.

US Government grants would provide guaranteed incentives for American universities to offer these kinds of courses, which would educate your populace and bolster the appeal of your higher education institutions to potential overseas students. The latter point would lead to many knock-on economic benefits (e.g. immigrant students deciding to stay in the US).

(oh, but rabble rabble gubment rabble bad rabble rabble because this is HN where libertarian capitalists roam free)

Government subsidized? You mean the library? Because IIRC the courses retail at a fortune (but are always on sale.)

It's not really. The most useful thing a teacher can do is give feedback. That said, one student among 300 doesn't get a lot of feedback.

Research for the US Military was/is an important source of funding. To be clear, it's not like the government is getting nothing in return.

Gifts from successful alums are also a big source now: http://facts.stanford.edu/finances.html

This sort of thinking is supported also by the fact that the number of views is not strictly decreasing.

I think it's safe to say at least 3/4 will not do any work. Another 3/4 of the remaining will drop out after trying, and another 3/4 will not understand a thing.

900 is still a decently large university class.

I don't think you belong here. You're supposed to whine and bitch bitterly about "Kids these days".

I figure the future of education looks like this. Where anyone can sign up for courses, and if they don't get it, they can either drop it or hire a tutor.

1 teacher per subject per 300 students is an anachronism. It should have gone away when textbooks became ridiculously cheap to print.

I think you're being way too optimistic. I'm guessing about 5% of sign-ups will ever actually follow the course once it starts, and of those, I'm guessing less than 5% of those people will complete the course.

Do you think they'll makes the numbers public?

Most likely NYT updated it. Easy mistake for someone not familiar with this topic to make.

that's what they said about financial experts ...

Uh, in "AI", the "I" is mostly just a hope or goal and not yet an accomplishment!

We have AI as a field mostly just because of long, aggressive funding by DARPA, early on especially of the MIT AI Lab. Since then DARPA has funded the prize where computer controlled cars chase across a desert. It appears that some people at DARPA are hoping that something good happens from their funding. Tough to say something good will never happen! If something okay leaks out of the research, then some lives of some US soldiers might be saved and, by now, may have already. In that case, good. DARPA is likely some of the very best money spent, even if for some years the work looks weak. At least with the money so far, a lot of quite bright people have TRIED to make progress in AI.

Some years ago, I was one of three researchers in an AI group that did some of the world's best AI work. I had two conclusions: (1) For the practical progress desired now, just use 'traditional' approaches of (A) see problem, (B) analyze problem, (C) find assumptions can make about the problem, (D) do some math starting with the assumptions, (E) take the conclusions of the math as progress on a solution, (F) use the best of what have to wrap up the best solution can see. (2) For the AI part, so far "We dance 'round in a ring and suppose while the secret sits in the middle and knows.".

You can use TunesViewer [1] to view iTunesU media and podcasts in Linux.

[1] http://tunesviewer.sourceforge.net/

While I agree that iTunes is a very cumbersome piece of software and it is more difficult to use on Linux, the amalgamation of educational videos is the big step forward. The technology has been around, like you said, for a long time to post videos online - but what good is it if those videos are hosted on a thousand different websites? iTunes U aggregates them all into a platform that can easily be searched, as well as enforcing a minimum level of quality in order to provide a higher quality learning experience. Is Apple truly that magnanimous? Probably not. But it's an ad-hominem fallacy to dismiss the quality of iTunesU on the basis of it's creator.

I see your point about not having access to the iTunes content on Linux. Fair point that should be addressed.

That said, I'm a big fan of iTunesU. I like the ability to download courses on my iOS devices and listening to the lectures on trains or at the gym. Also, they must save a lot on bandwidth by letting people download the videos (I go through the lectures a couple of times at least).

iTunes U (as in U for University) is a section of iTunes where you will find a whole host of University courses to download and watch. I mention this to many people and there seems to be very little awareness that all this high quality content is available on iTunes.

There are some great courses available (most for free), in fact I have just finished the Stanford iPhone programming course by Paul Hegarty which I highly recommend.

"Certainly nothing beats a good book, but often a good lecturer can give some intuition or motivations that are not customarily written down."

You are correct on both points. Early in my learning, I found both points to be important. And, for a beginning student, both points can be crucial: Without the second point, it is far too easy for a student to get into some not very good material or into some good material but, still, lost.

Eventually I got away from your second point.

Still, there is a version of your second point that lasts: For research, seminars and conferences are good, fast ways to keep up, see the forest for the trees, pick new research directions, etc.

When my company is successful and I retire, I will return to mathematical physics and, maybe, attend research seminars in, say, Boston.

Is Boyd still talking about convexity?

Convexity is an important concept, and the world is just awash in rock solid treatments.

So, for some positive integer n, consider n-dimensional Euclidean space. Suppose C is a closed, convex subset and point x in the space is not in C. Then there exists a closed half space that covers C and omits x. The boundary of this closed half space can share a point with the boundary of the closed half space so that the plane of the half space is 'supporting' for C. So, this is a 'separation' result. Yes, it generalizes past finite dimensional Euclidean spaces! Yes, it's a darned useful theorem! E.g., can ask for the point in C closest to x -- the point has to exist since C is closed. Then the plane through this point and perpendicular to the line to x is supporting for C. So, can get a nice projection result and use it for optimization and best approximation.

Okay, my point: This result is one of the most important about convexity but is standard in 'analysis'. I first saw the result in the 'advanced calculus' book, Fleming, 'Functions of Several Variables'.

Bigger point: Convexity pops up frequently in analysis. E.g., there is Jensen's inequality. With it can easily knock off a nice list of otherwise difficult to prove inequalities. E.g., in the L^p spaces, we can use Jensen's inequality to get Holder's and Minkowski's inequalities.

When we start with optimization, we should start with convex sets and also convex functions. E.g., there is a nice list of theorems of the alternative that are separation results for cones and polyhedra that can be used to establish some otherwise tricky results about duality in linear programming and are also key to the Kuhn-Tucker conditions in non-linear programming. Of course, the feasible region in linear programming is a convex set.

In non-linear programming maximization, there is a non-linear dual that is to minimize a convex function, and that can be the core of the powerful technique of Lagrangian relaxation.

When we have a norm on a vector space, the locus of all points distance 1 from the origin is convex.

In facility location we can be minimizing a convex function and doing so by constructing a sequence of planes supporting the hypergraph of the convex function.

Alas, I never heard a lecture from Boyd!

There are plenty of rock solid, highly polished books where the role of complexity is made clear!