I'm interested in learning about AI because I think it'd be cool to create something that is somewhat intelligent. That's what I'm hoping for. But I have a feeling that I'm going to end up getting a lot of abstract theory and only enough practical knowledge to create something basic like a recommender for related things on a website. So I'd really like to hear from other people who've done a masters or other specialization in AI:<p>What were you hoping for and what did you end up getting?
I have no degree on AI, or any sort of specialization, apart from studying by myself. But I do work with AI, using different approaches and techniques. /disclaimer<p>My view of AI is that it is as complex as it is fascinating. It might seem easy to just download an artificial neural network library, feed data to it and then see what it spits back, but what data should you feed it? And you want it learn? Or should you use a genetic algorithm instead of neural networks?<p>Those seem like easy questions to answer with a handful of math formulas, but the formulas will help you more on giving you a firmer assessment on the scope of the solution space, how to measure if it is really progressing, when the AI is in learning plateaus, etc.<p>But the big reward is in how to creatively use AI. By "seeing" data in a new way you can ask different questions, and might get some interesting answers.<p>I'm telling you all this because my view on how and when to use AI evolves constantly to this days, and I've been working with it for about 7 years.<p>So start studying right now if you really like this field. Not only will the formal AI education become easier, but you will probably be able to get more from it if you get there with some experience.
I have a Bachelor's degree in Computer Science, and I'm planning to complete my Master's degree in AI by September. Given what you just wrote I'm not sure following AI courses at a University would be best for you, since as you guess, there's lots of theory and not that much of practice.<p>In general, the practice you get is mostly related on the concepts -as in, implement this methods on abstract data and see how they work-, but the whole framework is setup so that you can in the end contribute to the advances in AI and develop new methods. If your goal is simply 'create something that is somewhat intelligent', I think you could easily read introductory books by yourself and implement existing methods. It would be a more efficient use of your time I believe.<p>In addition my particular Master was somewhat broad in scope, so that I learned something from very different fields (Computer Vision, Machine Learning, Decision Theoretic Control, Information Retrieval, Neural Networks..), but in the end I still don't feel specialized, and I'll probably (maybe not now, but after a couple of years' work) follow it up with a PHD.
I can't pretend I have an advanced degree, or any degree actually, but I did attend college, and I have looked at a number of different university curricula. And I have been paying attention to the obvious headlines in artificial intelligence and artificial general intelligence.<p>The courses that I saw and discussion that I have had with post-graduates in AI or related fields is very different from the topics that were being promoted by AI headlines or at AGI conferences.<p>If you really care about having the most relevant skills and information, in this age of Google and other online search tools, you need to do research yourself. You are almost guaranteed to NOT get the latest and most relevant information if you limit yourself to the courses listed in some particular university catalogue. Certainly the researchers at those schools don't.<p>If you do a masters and let your thesis be dictated by whatever your professors think is relevant or that you should pursue, unless your professor is really working on the leading edge of something, you are unlikely to be there either.
No masters or other degree but I've been interested in AI since the AI summer when LISP and Prolog were what the cool kids were wearing. At that point The pattern was already set. In 1962 a chess playing computer was an AI breakthrough. By 1987 I could buy a shrinkwrap chess game for my Amiga.<p>With time, today's AI comes to be seen as just clever programming and then ordinary programming like any other algorithm. 2012 saw computers which could beat any chess playing human in the world and the verdict was largely that success was due to how much hardware had been thrown at the problem not great AI.<p>That said, inference engines and logic programming and constraint networks are really fascinating and these days the tools and books are a web address and a modest download away.<p>Good luck.
I'm curious as to what others got out of a masters in Comp Sci. I didn't study comp sci as an undergrad and sometimes feel like there are concepts I'm completely ignorant of. I wonder if a masters in Comp Sci leads toward being able to handle complicated application development.