Learn about natural language processing.
- Communication cuts to the very heart of who we are as human beings. It's the language that we share that helps us understand larger concepts such as community, law, and justice. As human beings we're always trying to do a better job communicating. So it's not much of a surprise that we want our machines to do the same thing. In many ways, machines do a much better job communicating with each other than us as human beings. It's pretty easy to have two machines communicate.
There might be an occasional packet loss here and there, but when you send an email it usually arrives in its original form. Human beings, on the other hand, are always struggling to reach greater understanding. If you can deliver 5 or 10% of what you're intending, then you're a great communicator. The main challenge is that we can't communicate with machines in the same way that they communicate with each other. We're not like Neo in The Matrix. We don't have an uplink port that will allow us to connect directly to the network.
That means that the machines have to do a better job existing in our world. To do this, AI programs try to do something called natural language processing. This is when you can interact with the machine using your own natural language. We're all familiar with how to communicate with a search engine like Google. There's a little box, and then you type in different questions or phrases. You can type something like, "recipe for Belgian waffles." Then the search engine will match your phrase to popular results.
It will look through common recipe sites for the term "Belgian waffles and recipe." Natural language processing makes this interaction much more human. Imagine if you could say something like, "I'm cooking breakfast. "Can you give me a good recipe for those "big fluffy waffles?" even with this simple request, you have a lot of natural language processing. The machine has to understand that good is relative, so in this case the person is probably looking for the top recipes.
The machine also has to figure out what's a big fluffy waffle. It's pretty common for human beings to describe thing by their attributes. It would be almost impossible to come up with an AI program using symbolic reasoning to do this level of natural language processing. How could you come up with an expert that could record all the relationships between different words and phrases? You wouldn't want an expert in a room hand coding different ways to describe Belgian waffles.
Again, a lot of the work in this area has been using machine learning and artificial neural networks. Any time you send a text or an email, it potentially goes through servers that can process parts of your conversations. They don't usually do this because they're interested in what you're saying. Instead they do this because they're interested in how you're saying it. It makes sense that organizations that are interested in artificial intelligence also offer many free communication services. Google has access to anonymized versions of your email and voicemail to pick up how you have conversations.
Apple offers iMessage and Microsoft has Skype. These services give their AI programs a treasure trove of different types of human communications. They can use machine learning to see patterns in how humans use their natural language. But natural language processing isn't just about understanding the words. It's also about understanding the context and meaning. A few years ago one of the top Google searches was "What is love?" At the time, when you put that search into Google, you would get all long list of results.
Most of them were about biological pairing rituals and the importance of feeling connected. This was the kind of response you'd except from a network that's just matching keywords. Natural language processing gives machines the ability to better understand the larger world. If you're typing in "What is love?" into a search engine, then you're probably much more interested in romantic notions of love, perhaps even some poetry or insights into what it's like to be in love. You might just want to hear a hit song by Haddaway.
Human beings have written on love from the beginning of language, so there's sure to be plenty to see on the topic.
This course will introduce you to some of the key concepts behind artificial intelligence, including the differences between "strong" and "weak" AI. You'll see how AI has created questions around what it means to be intelligent and how much trust we should put in machines. Instructor Doug Rose explains the different approaches to AI, including machine learning and deep learning, and the practical uses for new AI-enhanced technologies. Plus, learn how to integrate AI with other technology, such as big data, and avoid some common pitfalls associated with programming AI.
- The history of AI
- Machine learning
- Technical approaches to AI
- AI in robotics
- Integrating AI with big data
- Avoiding pitfalls