In a little over one month, a little-known Artificial Intelligence (AI) course has skyrocketed in popularity, since the two professors announced that anyone anywhere in the world could take their course. Three days after the announcement, 10,000 people signed up, and a month later, the number has increased tenfold. The Stanford University School of Engineering has decided not just to show videos of the classes – a practice which has been going on for around a decade in some of the most reputable universities in the world (especially in America) – but to allow students to take tests, watch lectures, and participate in class discussions. As said in a recent article from SingularityHub “Classes of 1 million or tens of millions may be in our future. If Stanford can succeed in teaching classes of 100k+ students at a time, what will it mean for education in general?”
Who are they?
Sebastian Thrun is a robots expert, and researches machine learning and “probabilistic methods” – which was the subject of a book he co-wrote, “Probabilistic Robotics.” Peter Norvig, who used to work for NASA, is now the Director of Research at Google. Norvig co-wrote the book “Artificial Intelligence: A Modern Approach” – in which Thrun contributed a chapter – which is used in over 1200 universities in 100 countries. Clearly, these guys know their artificial intelligence.
Thrun also appeared on the TED stage, where he talked about his dreams of building a driverless car, motivated by a car crash that killed his friend.
Where did the idea come from?
For the record, this has all been done before – at Stanford, no less. Andrew Ng, the Director of the Stanford Artificial Intelligence Lab, was the first Stanford professor to offer such an online experience in 2008, also posting his lectures on YouTube, and automatically grading tests and homework. The course was called “Machine Learning,” and videos are still up on YouTube as well as iTunes, Vyew, WMV Torrent, and MP4 Torrent formats, not to mention full transcripts of his classes. This year, Ng is again offering his Machine Learning course for anyone for free, from October 10 to Dec 16 – Yikes! That’s the same time as Thrun & Norvig’s class! Well Ng’s class doesn’t have as many people signed up (“only” about 34,000), but if you want to register, click here. In fact, Stanford boasts a wide variety of courses equipped with full video lectures on topics ranging from robotics, to iPhone app programming, to programming paradigms.
In an interview with the New York Times, Thrun and Norvig said that Salman Khan – who started the absolutely gargantuan video-based-education website “Khan Academy” (not for profit) – inspired them. Khan has made countless videos (the amount of hours of quality video-education is staggering) on various topics such as math, history, and biology. But in the same year as Ng’s Machine Learning class, professors Sudeshna Sarkar and Anupam Basu offered their own Introduction to AI at the Indian Institute of Technology, which can still be seen on YouTube today. In fact, for anyone who wants to take Thrun and Norvig’s course, this might be a good video series to view as a functional primer. In fact, another video series from the same institute on the same topic is offered by a different professor (P. Dasgupta) here.
How does the Class work?
The professors warn that this will not be a cakewalk. I’m referring to the Introductory AI course, but note that Ng’s class (and a third Stanford course on Database) is run in the same way. Students are expected to spend at least 10 hours per week studying. That doesn’t include the 8 homework assignments and 2 exams (mid-term and final).
Textbooks are not required (obviously the classes wouldn’t be “free” if they were) but I’m guessing the Stanford students are using them, and I’m sure it would help understanding the material. SingularityHub also mentioned that they would be offering the textbook for free, but I’m not sure whether not that’s true or not. It would certainly be nice to the over hundred-thousand people who wouldn’t otherwise have access to it. I, personally, already have a copy… so I’m fine either way!
The online course will go on for 10 weeks starting from October 10, but the in-class class will start two weeks earlier, on September 27. There will be two 75-minute lectures each non-exam week. The lectures will be split into 15-minute videos, to make it easier to watch. Each lecture week will include a homework assignment, and there will be an exam after each month of lectures. Though non-Stanford students will not be able to obtain university credit, they will be given a certificate showing their grades relative to everyone else’s, and to just the Stanford students. Presumably, if you were in the top percentile, it may be a good thing to add on your resume.
There will be some type of forum set up for online students to rank questions that people have asked. By voting to filter out the less imperative questions, the professors can ensure that they give quality answers to the most highly ranked questions. There has been a webpage set up on Reddit for the online students to aggregate, share ideas, help each other study, and otherwise share their thoughts about it.
Homework will be graded automatically, though answers will not be given in the first few days, so as to deter people from cheating. The professors will allow six “late days” for assignments or tests, but essentially, you must submit your work on time to pass the course.
Update: It seems that the class – along with the website – is changing rapidly due to the massive numbers of people registering. They now include this on their website:
Due to its popularity, we will be offering this class in two tracks:
- the basic track – in which you watch lectures and answer basic quizzes. This is not the full course. But you will still learn all the basics of AI.
- the advanced track – is the full class, which aspires to be of Stanford difficulty. You do homework assignments and take exams.
Preparing for class
The one prerequisite they have on their class page is “A solid understanding of probability and linear algebra will be required.” In the interest of taking the course seriously, I have compiled some online resources that can be accessed taken at any time, including compilations of videos series, an actual course, and some text format information. This should help people catch up on those prerequisites:
- “Probability” from the Khan Academy. Yes! Salman Khan made 8 videos on explaining probability. But as part of his Probability course, he has a total of 25 videos.
- “Probability and Statistics” from the Open Learning Initiative (Sign up here). Open Learning Initiative is an amazing resource that not only provides free online classes, but allows instructors to create their own classes for students.
- “Introduction to probability” from Bowling Green State University. This isn’t a video series or a class. It’s just text-format (and sometimes pictures) that explain the very basics of probability. The only reason I’m including this is because it’s concise information, and so not overwhelming. There’s much more information on the internet that I didn’t include, because it gives people too much of a headache.
- “Linear Algebra” from the Khan Academy. Did I not tell you that he made a scary amount of videos? This course alone includes 50 videos. In fact, on the Khan Academy website, there are so many videos under linear-algebra that there’s practically no way you could watch all of them by October, but they’re all labeled appropriately so that specific topics can be covered when selected.
- “Introduction to Linear Algebra” from the University of Colorado. Unfortunately, these videos are all over an hour, much longer than the short chunks from Khan’s videos. But hey, if you have the time, it couldn’t hurt.
- “Linear Algebra” from the Michigan Institute of Technology. This class looks like it would be pretty helpful, and the videos are around 45 minutes.
All of those are free. And note that I didn’t include any courses that were only on statistics (I wanted my list to be geared more towards probability, like they said), but you can find tons of those courses online too.
Why does this class matter?
I believe that this class represents more than just an academic course. I think it represents a movement within the digital age. Universities may soon take this practice seriously, which may have ramifications for post-secondary education as a whole. And I doubt it would take much convincing for professors to get more into this – do you have any idea how much more famous those two professors are because of this? Might this gradually influence the tuition rates of universities and colleges by showing free content, reminiscent to how free websites are killing newspapers and magazines? Well, we have to remember that the part of going to school that’s expensive isn’t the education itself… it’s the certificate proving you got it.
Perhaps, at least, with the increased participation of such online classes, universities will find that getting this kind of press is beneficial to their careers, which could start a whole new education process. Maybe professors would also be judged more critically and be the subject of more scrutiny, as their lessons become increasingly broadcast around the world. On the other hand, it’s likely that professors will improve the quality of their class if they’re aware that they’ll be on a public record. Perhaps this type of online teaching will, in a decade or so, become a new milestone in a professor’s career, a new requirement for tenure, or a means of gauging a professor’s ability to teach.
Or, most frighteningly, this experience will have little effect on education, and universities will stop prioritizing such aspirations. I find that unlikely, but I simply don’t have an answer, and we can only speculate.
The bottom line is that these online classes are basically experiments. Sure, many of the tens of thousands will probably lose motivation, and performance is likely to decrease gradually as people give up. I honestly don’t expect that I will have the time to commit 10 hours per week, while maintaining a blog, a full-time job, and a social life. But never has a class grown to this size, and I have a feeling that the academic world will be watching this huge group of web-students.
So I want to take this opportunity to journey with almost a quarter of a million others (by the time the class starts, it may be even higher… after all, more than 7000 people registered since I began writing this blog post!), and together, learn about what so many of us strive to better understand: science.
Register to the Artifical Intelligence class here: http://www.ai-class.com/