oisinthomasmorrin

This course can be found at Elements of AI.

In the last couple of days, after finishing the exams, I decided it was time to do and learn about the things I really like. First on the agenda was breaking back into the field of artificial intelligence; after perusing a few courses (I had already completed Andrew Ng’s Machine Learning Course on Coursera), Elements of AI run by the University of Helsinki and Reaktor seemed like a good option.

The course is rather short: it is a 2 ECTS course provided by the university, which is further broken down into six sections. From the outset you are given the choice between doing each section week by week (this is encouraged by the providers as deadlines, they say, increase the likelihood of you completing the work) or at your leisure. As this was a refresher for me, I chose the latter. Interestingly, they say the course should take approximately 6-10 hours a week to complete – including extra reading and problem solving time, but by the end of the course, I found that it had taken a little over six hours in total!

To say the least, the course is very basic and very reader-friendly – which is what it was selling itself as, so that isn’t necessarily a criticism. It deals with everything on a high-level basis: the history of AI, AI as a word, the taxonomy of AI in relation to Computer Science and related sub-fields, simple minimax example, rudimentary Bayesian probability, stripped back linear and logistic regression examples, an introductory lesson into neural networks, deep neural networks, convoluted neural networks and adversarial neural networks, as well as more philosophical issues like the role of AI, the perception of AI and the future of AI. Throughout it avoided much of the technical jargon and had math that didn’t require much more than the ability to multiply two numbers and add. 

It is a course for the lay-person, a course with the sole purpose to demystify misconceptions and, if possible, whet the appetite of the course takers. The most rewarding parts of the course were those more philosophical sections that gave you scope to read more and engage and think about AI, as the mathematical and technical questions were contrived and much too easy – to the point that they were awkward – to really be considered engaging. The idea that your written work was peer-reviewed and that you had to peer-review the work of at least three other people, was novel and rewarding, as at times you were introduced to interesting and diverse opinions and ideas.

On completion of the course, the question I have to ask myself is: was it worth doing and would I recommend others to do it? As a refresher course, it was most likely too simple – all breadth with no depth, but as an introduction to the world of AI, it was entertaining, well-curated and certainly worth the few hours it took, and I’d recommend it to anyone who any neophyte or lay-person as a solid and broad eye-opener to the scope of AI. Though, if you want a meatier introduction, I’d skip the introduction courses and instead do Andrew Ng’s Machine Learning Course on Coursera.