In the big industry trends, many ordinary programmers are beginning to turn artificial intelligence, robots are especially popular phone almost every day companies come looking for companies to conduct business negotiations, telephone proxy robot, robot phone OEM and so on ...
However, due to the higher relative to the threshold of artificial intelligence, for many it slightly more difficult. So here I recommend a method of learning, this method compared to other methods is relatively smooth, easy to implement method of learning, so that more ordinary people into the field of AI programmers.
First is the simple introduction of artificial intelligence it! AI also known as AI, many people see it as machine learning, this is just one-sided. In AI, the symbols and logic is the key to artificial intelligence, but AI is advancing rapidly, statistical machine learning to hold a dominant position in the blessing of big data. Remember that artificial intelligence is not equate with machine learning, so this must be clear, not clear you can go and see Chou wrote "Introduction to machine learning."
The first question: AI easy? The problem vary, and if an ordinary programmer never had a bit of difficulty is carried across, because you have to face a lot of hard data and parameter adjustment formula, want good learning discipline, We need to find unique ways.
Any discipline must persevere with the following learning methods will let you do more with less 
1. First of all about this field, and establish a comprehensive vision, cultivate enough interest, then start learning the basics of machine learning, choose a progressive approach here courses to learn, the best course of experiments can be enough combat.
2. lay the foundation for machine learning have adequate knowledge, you can use machine learning to solve a practical problem. Then the machine learning methods can still be treated as a black box to handle.
3. After practical experience, we can consider continuing to learn. At this time there are two options, or to continue deep learning machine learning. Deep learning is the most fiery of machine learning direction, some of which have been with the traditional machine learning are not the same, so you can learn alone. In addition to deep learning, machine learning, statistical learning method also includes practical, integrated learning. If the conditions are enough to learn both, some rules are common to both.
4. after learning, you already have a strong knowledge base, you can enter more difficult to combat. At this time there are two options, you can choose to see the industry's open-source projects, for the purpose to change the code to read the code; you can look at specific areas of academic papers, in order to solve the problem and want to send papers.
5. Whichever who need excellent knowledge, and strong coding ability, so it can investigate and exercise levels. After this later stage, can be said that the field of AI into the door.
In addition also recommend some books assisted learning
Deep learning (paper): 2015 papers published on Nature, written by Daniel three deep learning community, read the full papers, giving a strategic height and a list of small hills feeling, highly recommended. If you can only read a paper to understand the depth of learning, I recommend this post. This paper has the same name Chinese translation;
recommendation, UFLDL: DL a very good foundation course, also written by Andrew Ng. There are detailed derivation, there is translation, and a high quality of translation;
recommendation, Neural networks and deep learning: author of the book is very good at plain language to express profound truth, though not translated, but reading is not difficult;
Recommended , Recurrent Neural Networks: a combination of actual case RNN tell you what is, after completion of the entire course, will have an effect on how you RNN has a very clear understanding, and this effect, does not have even read a few relevant papers of;
These are mainly introductory want to help, but could not find the target programmer to write, because the smart industry in order to specialization, is not easy, and need to put the years of accumulation and learning, of course, learning is very boring and tedious only the interest is the best teacher, but also allows you to power down.