Daphne Koller of Stanford AI Lab on online education | Hacker News Reader

Daphne Koller of Stanford AI Lab on online education

nytimes.com

55 pointsrobrenaud13 comments

I think this type of learning will be considerably more effective in quantitative subjects than qualitative ones. Maths especially but essentially any subject that has 'right' answers lends its self well to this as content and tests can be automated algorithmically.

I'm curious however to see how this works in humanity type subjects that require more critical thinking and higher order reasoning. These seem to me to still require human interpretation in order to assess understanding.

Perhaps it's easier to conceptualize how to structure classes that are quantitative and that's where the technology will move first. Also, the people with the talent to create the software might have a bias toward the qualitative subjects. The later will be a much harder problem to solve, but any student who has experienced the feeling of being a few steps ahead or behind the rest of the class would benefit from blended learning in any subject.

I think the first step is to just put the content online, which is what Kahn Academy does. The next step is to make the content interactive and gamified, and also have a sandboxed playground to test new ideas in.

I think coding is the new literacy which is why I'm working on www.trybloc.com, to me that level of instant feedback and interactivity is where the real innovation is.

I'm sure we've all experienced the HR videos with the questions at the end of each section to make sure you didn't just leave it running while you watched The Big Bang theory. Do we really think kids are going to become more engaged if school is more like an HR training session?

I do agree that lectures make no sense at all (the were pioneered when books were extremely expensive to reproduce; books are now free to reproduce), but I think limiting ourselves to "lectures, only now not live so we don't have to pay teachers and with extra busywork tests" isn't going to revolutionize the field unless all we care about is cutting spending further.

The Stanford class videos are not at all like what you are referring to.

The difference between HR videos and lecture videos like what Stanford is doing now is huge - the people watching the HR videos don't want to be there. People that signed up for the Stanford classes are interested students, which makes all the difference in the world.

So true. Someone should look at learning tools and systems from the "customer" perspective. Are they already interested/willing to learn? How hard is it to motivate them?

I think we all had these moments when we really wanted to learn something. Than the medium of delivery is more or less the least important aspect of it.

I see what you're saying about the quiz questions every few minutes.

On the other hand, she's not arguing for limiting ourselves to video lectures. Her point (see page 2) is that class time is precious and shouldn't be wasted on lecturing but on problem solving, critical thinking, and discussion, and this I totally agree with. (Why is it taking institutions so long to figure this out?)

The best teachers practice differentiated learning, teaching to a variety of learning styles at once. This is essentially using technology to maximize the human element necessary in teaching and then facilitate groups of students who are working at their own pace to demonstrate learning. Keeps students engaged and holds them accountable.

I tried out the database class, but didn't follow through with it because I found myself just teaching myself through looking things up online anyways (i.e. W3C tutorials, free online textbooks, etc).

I didn't find the class itself terribly worthwhile, but I was impressed with the quizes/tests/assignemnts. They seemed like a great way to evaluate whether or not you've learned something.

It's hard to describe, but it seems like they really got this part right, at least based on my limited experience with the DB class.

The central concept in this which was not mentioned was gamification. Education needs to be gamified, ie. every small step about it needs to be quantified, the student needs to receive instant feedback, the "game designer/educator" needs to see the data and improve the game.

Somewhere along the line the education system came to be about competition, who scores the highest gets to brag he's the cleverest and somehow superior. Instead, it should be about gaining competency in a fun way. Doesn't matter whether students reach it at 15 or 25 or older. (separate issue, but the core competency everyone in society should have nowadays is the ability to program in python).

Gamification is a gimmick that can add value when the people are already engaged, but you can't make people interested in education because you can get points or win a badge. They had to want to be there in the first place.

Some fields lend themselves to domain experts who can write software making large leaps in productivity, but many do not. There are a lot of desirable skills I'd prefer the education system teach before Python. In the end, I think you want a society where anyone with the need or interest can go learn it on their own. This presents a huge educational challenge that we should try to fix as best we can, but this problem will never be completely solved. We should use our resources to improve things as best we can.

how about paying real money as points. money has been shown to motivate students to learn. gamification is not a gimmick, it is a rephrasing of an old idea - feedback. all animals operate on instant/constant feedback. it's only in humans that the ability to operate without this is present, and it's a mistake to force everyone, including to children to operate in this feedbackless manner.

With regards to python, the low hanging fruit out there is just a few lines/hundred lines of python away. If everyone out there thought and acted like a programmer, so much inefficiency would be squeezed out that it would be equivalent to another industrial revolution.

Reply on news.ycombinator.com