Alternatives to the Rescinded AI Course

The plan to create a mandatory artificial intelligence (AI) course at KAIST was recently rescinded. The KAIST Herald takes a moment to ponder whether the idea deserved a complete scrap. Could we make do with an improved plan to better prepare ourselves for the anticipated Fourth Industrial Revolution? Below are two alternatives to the rescinded AI mandatory course.

I am sorry. I am very sorry to hear that the new KAIST president’s ambitions have been walled off by the student community. I am not sorry, however, to see another ill-presented dream get rejected by healthily alert students exercising their own rights. The presentation of the idea of the mandatory artificial intelligence (AI) classes could not have provoked the public’s cautionary antennae further.

There are several reasons for that claim. First, the term “AI classes” is simply too broad. The term was coined in the 1950s to undergo several transformations in the depth and breadth of its own definition; millions of print and electronic materials still pour out and make it difficult to even point to the beginning of a curriculum; applications and implications range over every single field anyone could think of. The implementation of AI classes could have meant too many different things to too many different people in the KAIST community, to the point that there couldn’t possibly have been a positive consensus supporting it. Second, the “mandatory” part must have scared off a lot of people, including myself. We are talking about people who have heard things from the firsthand witnesses of mandatory English classes. Third, the design of the AI classes was questionable. How difficult would it be? What mix of theory and application would it consist of? How many credits would it offer? It is a well-known fact people reject the unknown, and these proposed AI classes have always been veiled until eventually the plan to establish those classes was annulled. The mere abduction of a buzzword and its insertion into a curriculum were actions far from satisfactory. In short, the rather dictatorial declaration of the mandatory AI classes was, at best, a politically unpopular choice.

It is worth noting, however, that in these reasons lie the very opportunities that could indeed prepare KAIST students for the near future without having to suffer doubtful stares and inquisitive gazes from the public.

First, abandon the “one-curriculum-fits-all” approach. These mandatory AI classes must not be one-sized. It will have to differ from major to major, in order to cater for the different needs in each field. Different AI theories and applications will certainly have varying usefulness across fields of study, so it is only just that the AI curriculum be diverse enough to accommodate the equally diverse fields interlocked with AI.

Second, for these AI courses to truly settle into KAIST curriculum, they must cease to be mandatory. A diverse enough mandatory course will easily collect audience from each major successfully, but it will hardly be anything more than an array of showcases on current trends and applications. For a genuinely immersive and engaging learning experience on the topic of artificial intelligence, there must exist some form of “price discrimination” in order to naturally repel those uninterested and bring in those who are. One easy way to do just that is to make the course optional. It is a myth that making the AI classes optional will fail. Look around you at the hundreds of other major courses, each of which offers an in-depth coverage of advanced topics offered only to those who are ready to commit.

Third, the next — if any — announcement of a possible plan to open a new course on AI must accompany a full-fledged design of the course. At least, the announcement will have to make it look like the course could be deployed any minute. Only then will students be able to break the buzzword spell and see the AI course for what it really is.