This episode of SciShow is brought to you by Emerson.

Visit Emerson.com/WeLoveSTEM to learn more. There are all kinds of things that connect to the internet these days, from smart refrigerators that tell you when your food is expired, to office lights you can turn off with your smartphone.

Together, this collection of smart devices is called the Internet of Things, and it’s changing how we run our homes, stores, and even restaurants. But someday it might be able to do a whole lot more. In the futuristic, perfectly-automated smart homes you see in comic books and sci-fi, the house basically runs itself.

It knows exactly what you need at any time, and there’s a virtual assistant that can answer questions and control appliances. We’re not quite there yet, but a future with smart buildings that can do all those things isn’t as far off as you might think. There are just a few main types of technology that we need to develop first, and the good news is, we’re well on our way to figuring them out.

For a real smart home, you’d need a system powerful enough to keep all your devices perfectly in-sync, communicating with each other, and aware of what’s happening all around the house. So, when your smart alarm clock goes off in the morning, it might turn on the lights in the living room, start a pot of coffee, and turn the TV on to your favorite channel. But the system would also need to be advanced enough to know if your roommate is already up and watching TV, so it wouldn’t change the channel.

A system like that would need some powerful hardware, and a little speaker on your kitchen counter, at the moment, isn’t going to cut it. But there’s a pretty good workaround. Most smart devices are already connected to a remote, more powerful central computer that takes care of the advanced data-processing stuff.

So you wouldn’t need a supercomputer at home to run all the complicated programs that would keep track of everything and figure out the best way to automate your house. The data would just be sent to that remote computer over the internet, which would process it and then send instructions back to your devices. But even with enough computer power for a connected system, you still need to design software capable of automating an entire house.

We have assistants on smartphones, but they’re not that great. They’re only built to understand certain types of commands, and they still get confused. A lot.

To make that perfect virtual butler, we’ll need advances in two fields of computer science: natural language processing and machine learning. Natural language processing is centered around designing programs that can understand us the way we naturally speak. That means your smart home would have to understand original sentences, their context, words with different meanings, and all of the other things that make up language but because language is so complicated, we probably won’t have computers that can do that perfectly for quite a while.

One thing that could help, though, is machine learning, where computers gather data and teach themselves. So instead of programming a billion words and grammar rules into a smart home, the software could just teach itself what to do. Devices all over the world would collect data on the commands people gave them and then adapt their programming and algorithms in response.

You probably already have an assistant on your phone that uses a version of this. That’s how it learns about things like new slang or fashion trends. We just, of course, have not gotten to the point where they can learn to understand us perfectly.

And, honestly… I don’t understand what people are talking about these days sometimes. With more advanced machine learning, your home could figure out how to answer every question you could possibly ask it. Based on the data it collected, it could just learn to only suggest dinner recipes after 5PM or only brew decaf coffee after noon.

So if you’ve ever wanted a virtual butler waiting on you hand and foot, there’s still hope. Finally, and maybe most importantly, your smart home would need to be totally secure. Besides having routers or other devices to protect your WiFi network from snoopers, the devices in your home would need to be able to shut down attacks.

Everything connected to the internet would be an access point for the rest of your wireless network. So hackers could use something like a smart fish tank to get to the rest of your accounts and passwords. Which sounds ridiculous, but that’s exactly what happened to a casino last summer.

A perfect smart home would need to be protected against both current attacks and any vulnerabilities that are discovered after the appliances are built. Not all companies think about that when they’re building something small, like a fish tank. So everyone would have to get on board.

And your house would still need to protect itself even if the company that made your devices went out of business or stopped updating its software. Like, you shouldn’t need to buy a brand-new fridge after a year because the startup that built it went under and there’s a new virus going around. To get that perfect smart home security, we’ll need to figure out ways to keep updating all of our appliances, no matter how small, and to make sure those updates are sustainable in the long run.

Perfect smart buildings will transform our lives someday. From how we live to how we shop to how we do our jobs. But there are a lot of developments in computer science to be made before that can happen.

So for now, our technology isn’t quite perfect yet. But we’re getting there. Thanks for watching this episode of SciShow, and thanks to Emerson for sponsoring it.

If you want to support us, check out Emerson.com/WeLoveSTEM to learn a little bit more about what Emerson’s about.