Stanford AI Class - Mid Term Report
Written by Sue Gee   
Friday, 25 November 2011
Article Index
Stanford AI Class - Mid Term Report
Learning - students and tutors alike

Sue's take - the learning experience

I class myself as an persistent student - I've been a full time "bricks and mortar university student, a part time face-to-face student, a distance learning student and so joining this online class was gives me another new educational experience. 

One strong impression from it is that it requires more time and effort than I'd anticipated and I wasn't as well prepared as I could have been. Yes, I armed myself with the textbook (A Modern Introduction to Artificial Intelligence co-authored by Peter Norvig) and started to read it - but I would have better to have brushed up on probability and formal logic.

It was only after class started that I noticed links to suitable preparatory material (I don't know if they'd been there before) but by then keeping up with the course was proving to be a problem. So much so that I fell behind in the very first week and missed even starting the homework. So after weeks of eager anticipation I felt crushed.

However, despite the set back I was really enjoying it and wasn't going to give up.

Then in week two the difficulty level, for me at least, really ramped up. In his introduction to Unit Three Sebastian told us it would cover structured probabilities using Bayes networks and warned us to expect something pretty challenging - it was and not always for the right reasons.

For example not only was some of the terminology new but, having never done logic, I was finding it difficult to interpret the symbols appearing on the screen. These are more fully explained by Peter in the next unit. If only they could have been presented the other way around!




Ramping up the level early in the course is a well known teaching technique - it enables the students who make it through the ordeal to be more resilient going forward and it worked for me!

Another technique I'm familiar with is posing questions that anticipate what is going to come in the next part of the course. In some ways this is more fair in a face-to-face situation where the students can ask for clarification of the tutors. However, in an office hours video Peter defends the practice of using quizzes in this way explaining that such question were designed to make students to think more deeply about what they are learning - and that's probably the answer he would have given face-to-face.

One of the things I'd looked forward to was the opportunity to join in discussions with fellow students and with the tutors. Although the two external forums on reddit and aiqus were doing an excellent job it was disappointing that there wasn't a single "internal" discussion forum. However that omission has been rectified and now the Discussion tab is where students get to suggest, and vote on, questions for the tutors to answer during office hours.




After the failed experiment using Google + Hangout, Office Hours in which Peter and Sebastian answer the most popular question  have been done by video.

It's not instantly-interactive but the comment stream below does add an element of interchange of ideas. The topics that are tackled are expansive rather than nit-picking which makes the session well worth finding time (around 15 minutes) to watch.




This is a course in which its not just the students who are working hard. So are the tutors are they are also experiencing a earning curve. Asked in an office hour question about whether the experience of teaching so many students online would change their future teaching methods Peter commented:

We're doing our best to keep up day-to-day. The impression from the quizzes and homework tells that with a thousand times as many students there's a thousand times more way to misunderstand and get some assumptions wrong.

And presumably just as technical hitches at the student end means that it takes longer than an hour to watch the videos, it takes more than the allotted time to record them too.

However, as the course has progressed, there's been a very noticeable improvement in the flow of announcements, corrections and pointers to resources, including to aiqus, which has became adopted as the official place to discuss the units, the homeworks and the midterm, complete with excellent moderation.

Additional feedback has helped. I'd been very disheartened by the 0% for my first homework so the arrival of an email which informed me "as you probably know" that it wouldn't drag down my final score - the lowest two homework scores would be knocked out. That was a great boost!

And while on the topic of explicitly telling students things it's assumed they already understand, it would be good some help with the user interface. Like many students I'd been re-running videos to reach the questions (and the marked questions in the case of completed homeworks) until it was pointed out in a comment stream that clicking on the black ? in the interface takes you straight there! That would have save me hours!




Hopefully the course will have another run as a free online course (a comment from Peter and Sebastian in an office hours video suggest they intend to re-run it) and they will take on board all the feedback received.

For myself I'm hooked and the fact that Stanford has announced nine free online computer science courses for the coming semester just means I'm spoilt for choice.

I-Programmer's verdict

So there is a lot to niggle about, but at the end of the day is it all worth it?

The answer is a resounding "yes".

None of our three students would have missed the opportunity to encounter the ideas and test the fact that they have learned it correctly.

Some aspects of the course have been frustrating but overall perfectly forgivable in such a huge undertaking. It is very easy to make a list of what is wrong with the course but at the end of the day it seems to work and we see it improving week-by-week. The forums have been a buzz with questions and lots of friendly helpful clarification.

If nothing else it proves the demand for this sort of thing and to have so many people learning about something as complex as AI because they want to simply makes the world a better place to be.





To be informed about new articles on I Programmer, subscribe to the RSS feed, follow us on Google+, Twitter or Facebook or sign up for our weekly newsletter.

Last Updated ( Friday, 15 June 2012 )