Eighth report from the Norvig/Thrun/Stanford/Know Labs Artificial Intelligence course

(Other posts tagged ai-course.)

Here is my eighth participant's report from the Stanford Introduction to Artificial Intelligence course.

Last week's report focused mainly on the midterm exam, and on a conversation with Sebastian Thrun. This report conments on the midterm exam and reflects on the way the course has run this week.

First, some comments on the midterm exam

1. According to information just received from Sebastian Thrun, 23k students passed the midterm, with 85% currently falling into the B+ range.

2. There was a surprising amount of post-exam discussion between students about the substance of the exam. There was some whingeing, but not much given the very large numbers on the course.  The interesting discussion fell into three categories:

• learner support - as in student A expressing anxiety having got an 85% mark and asking the world if he should give up, and students B, C, and D chipping in with supportive observations based on their own performance, their own attitude to the opportunity presented by the course, why student A should hang in there.
• learning support - as in a discussion developing about the subject matter of a particular middterm examination question, with links to relevant resources, contrary points of view, and so on.
• process-related questioning - as in students wanting to know the overall distribution of scores, numbers completing the middterm etc.

3. None of this is surprising, and there is obviously plenty of scope for a Hawthorne effect; but what seems clear is that with a large enough number of participants, and with the right discussion support tools in place (for example Aiqus, with very light touch moderation), the sense of being a lone learner is largely overcome, without the course itself needing to have designed into its activities opportunities for learner-learner dialogue.

Second some comments on this week's study

4.   Each of this week's three units - on Games, Game Theory, and Advanced Planning - have been taught by Peter Norgiv. Correction: all three units were made up of videos featuring Peter Norvig. (The correction itself emphasises the wholly odd nature of this course from the point of view of the student....) The three units have been hard work and the associated homework assignment felt demanding and was time-consuming.

5.  The density of feedback quizzes for these three units been noticeably higher than for the previous Norvig units; and this is a good thing, because it gives students much more of a sense of being personally tutored. (My fifth report has more to say about this.)

6. The materials varied greatly in their level of difficulty. This has been a constant aspect of the course to date. I feel neutral about it; but it does sometimes have the effect of making you look for more complexity than is actually there.

7. The section of the homework on particle filters (see below for Sebastian Thrun's overview) contained several questions that required a lot of study outside the content already covered (an in my case a possibly vain attempt to answer a question from first principles). This is no bad thing, but this is the first time on the course when it has not been possible to tackle a question on the basis of ground already covered.

Finally, two overall observations

8. I do not know what the "credit point equivalent" of this AI course is to a UK HE course, but I suspect that it is, say, a "20 Credit Unit". So far, on the KnowIt YouTube channel supporting this one course, there are now nearly 700 short videos. By the time the whole course has been uploaded, there will be close to 1000. For those entertaining this kind of  course design: be warned! A large number of assets will need to be managed and updated. On the other hand, with so much "atomisation" of content, the scope for amendment and improvement - withouth having to change large items of content - is  considerable.

9. With two thirds of the course completed, what would really help now is some stock-taking that focuses on how the different areas covered and coming up next fit together, possibly with some kind of diagramatic representation.