The ‘Internet of Things’ Faces Practical and Ethical Challenges

Photo illustration by Julia Schmalz

It’s a weekday afternoon at Carnegie Mellon University, and Anind K. Dey wants another cup of coffee. But Mr. Dey, director of the Human-Computer Interaction Institute there, has a meeting in 15 minutes. So he asks the Amazon Echo sitting on his desk: Do I have time to get coffee before my meeting?

The Amazon Echo, a “smart speaker” that can provide information and control other nearby devices through voice interaction, checks Mr. Dey’s calendar to see what time his meeting is and who will attend. It then communicates with the smartphones of the meeting attendees to detect where they are and what method of transportation they’re using to get there.

The speaker also communicates with a sensor in Mr. Dey’s favorite cafe to check how long the line is, and factors in how long it’ll take him to walk there. Finally, the device will reach a verdict on whether he has time to grab a coffee.

That’s how the Internet of Things project underway at Carnegie Mellon looks on the ground. In July 2015 the university announced $500,000 in seed funding from Google to begin the effort. Mr. Dey, the lead investigator of the project, envisioned it as a way to make life more convenient for students, faculty, and staff, and to collect helpful data about the university’s campus. Google also provided Mr. Dey and his team with access to its proprietary technologies.

Next: The Innovation Issue

Nervousness over the economy and questions about the value of a college degree have contributed to growing expectations that colleges must make career services a priority. This special report on innovation examines some of the career-counseling efforts underway — by colleges, start-ups, and collaborations between the two. See the entire issue here.

• When the Teaching Assistant Is a Robot
• Reinventing the Career Center
• Shadow Those Students, for Their Own Good
• It’s Time to Change What We Mean by ‘Credential’

The IoT, as it’s known, works through a network of internet-connected devices, such as wireless sensors and smart products like phones, speakers, tablets, and watches. The sensors, many of which are about half the size of an iPhone’s screen, can be placed virtually anywhere — including on toasters, coffee makers, doors, windows, and walls.

“You can start to get answers to questions that would’ve taken a fairly significant effort to figure out by yourself,” Mr. Dey says. For instance: Why is my office so cold? Is my colleague in her office right now?

The IoT has already been incorporated into entire cities, including Songdo, South Korea, where street lamps adjust their brightness according to the number of pedestrians in the area. That’s not yet the reality at Carnegie Mellon, though thousands of sensors have been placed across the campus over the past year and a half, Mr. Dey says.

To start, the researchers purchased hundreds of sensors that transformed their offices and labs into smart spaces. “We started putting sensors everywhere because we didn’t know what to put them on,” he says. “So we said, Let’s not try to predict what people want to know.”

When the Internet of Things project was first announced, Mr. Dey told The Chronicle to imagine a scenario. You’re driving to the campus, and your car tells you in advance that one parking lot is full and advises you to park in a different one. The technology exists to make such a world possible at Carnegie Mellon, Mr. Dey now says, but that sort of scenario isn’t playing out yet.

One key limitation: The sensors Mr. Dey and his team initially purchased were battery powered. When he put sensors in his office, he was changing the batteries every two days. “I’m willing to do that because I’m trying to learn something,” he says, but most people on the campus won’t be willing.

Though the project is still in early stages at Carnegie Mellon, it’s already proved useful for some on the campus. Beyond coffee, you can use the system to print more easily. Take a photo of a printer through an app, and the printer will then communicate with your smartphone to produce your documents.

Soon, university researchers hope to have a system in place for visually impaired people that uses Bluetooth Low Energy beacons, which can be used for indoor location-tracking. The system will help the visually impaired find the easiest way — with the fewest staircases and obstacles — to get from one building to another.

Along with enthusiasm, the concept of the Internet of Things has drawn criticism from cybersecurity experts and others for the privacy concerns it raises. “People have a justified worry about their every move being marked in some database,” says Frank Pasquale, a professor at the University of Maryland’s Francis King Carey School of Law. Many big-data projects are deployed on a large scale before researchers thoroughly weigh the benefits and costs, says Mr. Pasquale, who studies the intersection of technology and the law.

The IoT can be fruitful territory for hackers. In a 2014 experiment, a group of researchers at the University of Michigan hacked into IoT infrastructure in a small town in the state and seized control of nearly 100 traffic lights.

Worried observers ask: How much information about me is being collected? Who has access to it? What is it being used for?

Look at Oral Roberts University, Mr. Pasquale says. The Christian institution attracted attention recently for a new program that requires students to record their physical activity with the fitness trackers known as Fitbits. While officials there said they were trying to encourage healthy habits and see if there was a correlation between exercise and academic success, critics contended that the university could use the devices to track students’ locations and other sensitive information.

The more a campus is connected through such devices, Mr. Pasquale says, the greater the potential for the system to be abused. And it’s not just privacy that’s a concern, he says.

Say a college uses data collected through an IoT system like Carnegie Mellon’s to determine whether a student is academically at risk. If the data reflect certain things about a student, that student may be flagged. Mr. Pasquale is worried that students won’t be able to challenge the data and the algorithms used to interpret them — even when advisers use that information to make important decisions about students’ college careers.

Mr. Dey acknowledges that, when it comes to data, he and his team are “collecting everything we’re allowed to collect.” When the IoT is trying to answer questions and communicate information about people and places, that inevitably involves sensitive data, he says. “At times, we’re being extremely invasive.”

At Carnegie Mellon, Mr. Dey stresses, the IoT system is storing data from private spaces, like offices, only when people have opted in. In public places, researchers are being extra careful, he says. In the cafe where Mr. Dey likes to get coffee, he had originally planned to install a sensor with a camera, but “that just leads to a lot of uncomfortable questions,” he says. Instead he decided to use a smaller sensor that would only measure the length of the line.

A group of researchers is developing privacy tools. “In the Internet of Things, you’re just entering different spaces and interacting with different devices, and you may not be aware that sensors are present, let alone what their privacy policies might be,” says Norman M. Sadeh, a professor of computer science at Carnegie Mellon and one of the privacy researchers.

He says his team has created registries that enable people’s smartphones to discover whether there are sensors in an area and to understand what data are being collected. In certain spaces, people can request that data about them not be collected. For instance, if the sensor uses facial-recognition technology, people can have their faces obscured from the sensor.

Sometimes you can’t opt out, Mr. Sadeh says, though he points out that the advance notice will allow people to choose not to enter a room, or to turn off any smart devices before entering. Still, for the most part, people have to take action to ensure their privacy.

The eventual goal is to minimize the number of physical sensors by teaching the IoT “how to reason about the real world” on its own, Mr. Dey says. Fewer sensors means fewer privacy concerns, he says.

Mr. Dey believes the IoT will soon appear on other college campuses, and will eventually include billions of devices and sensors. Smart homes, smart traffic signals, smart factories — the possibilities are endless, he says. The system’s effectiveness relies on that larger scale. “If my thermostat can only talk to my car, that wouldn’t be that exciting to me,” he said.

For that to happen, Mr. Dey says, the sensors have to become cheaper (the first ones purchased by Carnegie Mellon cost about $29 each), and people have to become educated about how the Internet of Things works. “Unless you know what questions to ask of Amazon Echo,” he says, “it’s just a box on the wall.”