That's a great question! There are a ton of uses for machine learning, and a lot of different tools and techniques that fall under that general category. You have a good insight about how people will often just think of it as self-aware computers, but there are really a lot of incremental steps between your cell phone and the fictional movie version of AI.
On the simple end of the scale, statistics are really powerful, and are at the foundation of most if not all machine learning techniques. Where I work at New Relic, Inc., we help people solve problems in their software, and our AI folks use simple statistics to learn and detect what is important for them to see to solve these problems. They're also working with some more advanced techniques to learn patterns and predict and analyze events relating to those computer systems.
I see new uses for machine learning all over. Just like computers went from being special tools used by only a few people, to being in people's pockets, cars, microwaves, and everywhere else, machine learning is become more a part of our lives every day. Many of the useful features of the apps and sites we use daily are driven by it.
AI techniques are used all over the internet to learn what people want to see and will respond to. These tools can be used to help accomplish most any goals, so the ethics of uses for machine learning has been an increasing topic. While concerns of a SkyNet are likely very far away, machine learning is an increasingly powerful tool, and thinking about what we want to accomplish with it for our societies is very important.
If you are interested in getting into learning about machine learning and AI, there are a bunch of great, free resources online. Coursera has a lot: https://www.coursera.org/courses?query=introduction%20to%20machine%20learning It can feel intimidating at first, if you are new the math. It did for me :) Learning some basic statistics can help you understanding the concepts. But just getting a general idea from real-world examples and visuals is great, even if the math is more than you get at first.