Artificial intelligence in homes
Some say it’s only a matter of time before your home becomes smarter than you. Whether that’s a good or a bad thing remains to be seen. But certainly, there will be times when having a home that is smart will come in handy.
Imagine you’re in the grocery store and can’t remember if you’re out of milk and need to buy more. You pull out your smartphone and after a few finger swipes, you’re looking inside your fridge. The Samsung Family Hub refrigerator will do that. That is, of course, unless you have the LG ThinQ fridge, which would have already made you a shopping list based not only on what you have in the fridge but also, which items are about to expire. The ThinQapp also works on your phone. It supposedly can read your facial expression and tell if you’re getting drowsy behind the wheel and try to wake you up. Some cars already can take over the driving for you.
robots with AI
If you do get grumpy the cheerful CLOi robot might be able to brighten your mood. CLOi is a digital assistant (not unlike Amazon’s Alexa or Apple’s Siri) that integrates LG’s whole suite of home appliances so it can help you decide what to make for dinner based on what’s in the fridge and then preheat the oven to get you started.
Going beyond the kitchen, other areas of the home that are getting an education include the laundry room (sensing the type of load and learning the user’s schedule), the family room (smart speakers like Amazon echo or Google home, and smart TV’s with voice command and smart search), the nursery (smart baby monitors) and the entrances (smart locks and smart Wi-Fi enabled security cameras). Then there are also robotic vacuum cleaners like the Roomba, the Eufy and others that can clean the floors even if you’re not home.
Technology Secrets
Behind these advances are a combination of three technologies that are now affordable: sensors, connectivity, and artificial intelligence. The price of sensors benefited substantially from the huge volumes produced for smartphones. Connectivity is also exploding thanks to the Internet of Things, which will become even faster and more capable with the launch of 5G wireless. The main advance with 5G (5th generation) wireless besides its higher speed and lower latency, will be its ability to connect many more devices at once. This will be critical for the kind of technological world we are moving into, with billions of connected devices, including millions of cars that will (eventually) be talking to one another.
Of all these, artificial intelligence, or AI, is perhaps the least understood. We often hear the term AI used interchangeably with smart machines, big data, or machine learning. Are these all the same? Yes and no. They are not exactly the same, though they are closely related.
Artificial Intelligence in the home
The term artificial intelligence was coined in 1956 by the American computer scientist John McCarthy. He said “making a machine behave in ways that would be called intelligent if a human were so behaving”. McCarthy also said that “every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it”. That statement has yet to be proven true in practice. While a skilled programmer can get a computer to simulate any aspect of thinking or learning that can be precisely described, it’s the precise description of how the human mind works that has proven elusive to this point. Still, the number of functions that have been largely represent the state of the art of AI today.
Two AIs
Artificial Intelligence is divided into two categories: applied AI, sometimes called narrow AI, and generalized AI. As the name suggests, applied AI is what is being used today to power your smart speaker. It also drives the recommendation engine on Amazon and is what controls self-driving cars. One thing that distinguishes AI from previous types of computing, is its ability to recognize and respond to ordinary language. In general, narrow AI is a type of computing that has been developed to perform a single task or a set of tasks, which could be quite complex, such as driving, but is still limited to that task. Generalized AI would be the more human-like intelligence which is not here yet and may or may not yet ever be. (A good discussion of that question can be found in the book The Fourth Age, by Byron Reese).
About 60% of today’s AI, has been developed using machine learning. Machine learning is a way of getting a computer to perform certain functions, without explicitly programming it. It’s a lot like learning by memorization using flash cards, rather than by understanding. If you showed a computer ten thousand orange objects, along with the word “orange,” then another ten thousand objects that were not orange, with the words “not orange”, that computer would then be able to recognize orange. Considering that this can theoretically be done in seconds, a machine trained in this way can learn many things. It can also continue to “learn” as it receives new data, so long as the data is categorized. The rest of AI mostly consists of rule-based systems built on “if-you-see-this, then do-that” type rules which must be painstakingly encoded by programmers. Each is useful for different types of applications.
It’s all about speed
One reason that this type of machine learning has become so popular is the availability of Big Data. While the idea of programming computers this way would have at one time been considered extremely tedious, the availability of high speed internet, the plummeting cost of cloud storage, and the incredibly fast speeds of today’s computers make this machine-learning approach far easier than having scientists and programmers trying to come up with mathematical equations for everything, which is how programming used to be done.
One downside of this type of computing, which is also known as neural or neural network computing, is that a computer can be extremely good a performing a task, say, picking out faces in a crowd. What it can’t do however, is tell you how it knows what it knows. There is no computer program that one can look at to gain insight. The knowledge is embedded in the complex network of associations, and cannot be extracted, or for that matter, verified, except through extensive testing.
You can expect to see more and more of this in the future. According to a report by Dell Technologies, 72% of business leaders expect AI to be “the biggest business advantage of the future”. That means there will lots of money invested in it.
So, will the devices in your home soon be smarter than you? Of course, it depends on how you define smart, but while each of these gadgets might be able to do one or two specific things better than you can, not only do you have countless skills, but you also possess an overriding knowledge of when to use each one and how to combine them as the situation warrants. It will certainly be a long time before any kind of machine can do that.