Keywords

1 Introduction

You get up and go into your smart bathroom to take a shower. Your shower is programmed to remember your shower habits, so you hop in and the water temperature, spray and lighting are just the way you programmed it. A smart home is defined as one that is connected, automated, and remote-controlled by the use of internet and mobile devices to help you control different features in your home [1].

If it were truly an intelligent home (i.e., AI-powered), when your alarm clock goes off in the morning, it would have scanned your schedule and turned the shower on to your preferred spray, lighting and temperature of 103 ℉. You haven’t been feeling all that great today, so when you get home there’s a package delivered by drone and you open it to find cold medicine. Your health sensors embedded in your bathroom detected signs of an illness and placed the order for you so it would be there when you got home.

The smart home market will approach 40 billion USD by 2020 [2] with household penetration expected to grow from 27.50% in 2019 to 47.40% in 2023 [3]. But connected smart home devices are only the beginning when it comes to smart home technology as the popularity grows and the technology is driven by artificial intelligence (AI).

Experts say that we will move from turning lights on and off by giving voice assistants commands to an intelligent home. Robotics will give us machines that offer a helping hand with cleaning, cooking and more. Central to this will be the data that the smart home collects, analyzes and acts upon, turning our gadgets into intelligent homes. Currently, AI-powered gadgets are voice assistants that get the latest news, weather forecast, or turn on our lights when we tell it to [4]. Even though these are “AI” systems they still continue to rely on human initiative because we tell them what to do or schedule them to do what we want [5].

Yet smart devices and homes come with well-described privacy and security issues. As homes become more intelligent or adaptive [6] in which the home seeks to adapt to its inhabitants and respond to their informational and comfort needs, do the inhabitants understand what that means or what makes a device or home smart?

Between February and June of 2019, we conducted semi-structured interviews of 40 smart home consumers to understand their perceptions of smart homes. We asked them questions about what makes devices and homes smart, purchase and general use questions, installation and troubleshooting questions, privacy and security questions, and safety questions (see Appendix A). For this study we were interested in what participants said makes the devices smart and what they said it means to have a smart home. This paper describes a subset of collected data centered on what participants thought the IoT means and makes a device and a home smart.

2 Background

For our AI assistants to assist us and make our decisions, machine learning technology enters into the picture. According to Forbes [6], machine learning is an application of AI that enables a smart device to learn and improve based on its experience as well as data it collects and analyzes. In the context of our research we consider artificial intelligence as the development of computer systems able to perform tasks that normally require human intelligence and it is the ability of computer programs to think and learn. Our devices can do a lot for us and, in the future, they will have the ability to adapt to our needs and goals. But do consumers of smart home devices understand what makes a smart home smart or do they confuse smart with connected?

Just what is a smart home anyway? According to Consumer News and Business Channel (CNBC), a smart home is defined “as a home that is equipped with network-connected products for controlling, automating and optimizing functions such as temperature, lighting, security, safety or entertainment either remotely by a phone, tablet, computer or a separate system within the home itself” [7].

The first ‘wired homes’ (i.e., homes with interactive technology) were actually built by hobbyists in the early 1960s [8]. What makes a home smart is the interactive technologies that it contains. According to Aldrich [1] a “smart home” is defined as a residence that is equipped with computing and information technology that anticipates and responds to occupant’s needs, promoting comfort, convenience, security, and entertainment managed by technology with the home and connections to the world beyond.

In the homes of the future, computer software will play the role of an intelligent agent that perceives the state of the physical environment and residents using sensors and artificial intelligence techniques that will take actions to achieve specific goals for the people living in the home (e.g., minimizing the consumption of resources, or maintaining the health and safety of the home and residents) [9].

But little is known about what consumers believe makes a device or a home smart. PricewaterhouseCoopers (PwC) hired a global research firm to conduct a survey of 1000 respondents in October 2016 to understand consumers attitudes and experiences with the connected home, i.e., smart homes [10]. One of the questions they focused on was “to what degree do consumers understand the concept of IoT and the smart home?” Results showed that the term “internet of things” was largely unfamiliar, yet the idea of devices being smart is a concept most respondents understood. What is smart according to respondents is: the ability to collect/monitor data and project future usage for efficiency, ability to connect to other devices, or something that helps out with everyday life and makes it simpler.

Smart homes are still in their infancy and react mostly to commands, but the smart homes of the future will make intelligent decisions for us as they run our lives in our home environment. There are big differences and many implications between smart device consumers adopting individual devices and signing up for a smart home life.

3 Methodology

Between February and June of 2019, we conducted a semi-structured interview study to understand end users’ perceptions of, and experiences with, smart home devices from purchase decision to implementation to everyday usage (see Appendix A). This paper describes a subset of collected data centered on what is the IoT and what makes a device and a home smart.

3.1 Participant Recruitment

We hired a consumer research company to recruit general public consumers who had two or more different smart home devices. To determine eligibility, potential participants completed an online screening survey about their smart home devices, basic demographics, role with devices (e.g., purchaser, installer, troubleshooter, user), and number of household members. We reviewed the screening information and selected participants who met the minimal requirements. General public participants received a $75 prepaid gift card as compensation. Thirty-three general public consumers completed the interview and we identified seven participants from federal employment.

The study was approved by the National Institute of Standards and Technology (NIST) Human Subjects Protections Office (HSPO). Prior to data collection, participants were informed of the study purpose and how their data would be protected. Data were recorded without personal identifiers (instead using generic identifiers. e.g. P1, P2…) and not linked back to individuals.

3.2 Data Collection

We collected data from 40 semi-structured interviews that lasted 41 min on average as well as demographic information. The in-depth interviews afforded a greater richness of data, the ability to ask follow-up questions to explore, and the opportunity for participants to provide other relevant information not explicitly targeted by the interview protocol.

We piloted the interview protocol with four individuals to determine the face validity of the questions and language. We made minor adjustments to the interview instrument based on feedback from the pilot. We audio recorded all interviews and hired a transcription company to transcribe them. Interview questions addressed several areas in the following order: understanding of smart home terminology; purchase and general use; installation and troubleshooting; privacy; security; and safety. For this paper, we are focusing on the questions about what is the “Internet of Things”, what makes devices and what makes homes smart.

3.3 Participant Demographics and Devices

Twenty-two males (55%) and 18 females (45%) completed the study. The majority were between the ages of 30–49 (see Fig. 1). Participants were highly educated with 18 (45%) having a master’s degree or above and 20 (50%) with a bachelors’ (i.e., BA/BS) degree. (see Fig. 2). Of the 40 participants, 32 (80%) had installed and administered the devices, while eight (20%) were general device users. Participant occupations included diverse occupations from assistant principal to community arts specialist. Six participants had occupations in computer science or IT fields. Table 1 shows the participant occupations.

Fig. 1.
figure 1

Males & females by age category

Full size image
Fig. 2.
figure 2

Education

Full size image
Table 1. Occupations
Full size table

Smart devices in participant’s homes included: security cameras, motion detectors, door locks, and water-leak detectors. Smart entertainment devices included: smart televisions, speakers, streaming devices, and other connected media. Devices in the home environment category included: energy-saving technologies (e.g., smart plugs), lighting thermostats, and temperature sensors, smoke detectors, and air quality sensors. Smart appliances included: large and small connected appliances (e.g., refrigerators, coffee pots, robot vacuums). Virtual assistants included voice-controlled devices such as Amazon Alexa and Google home. See Fig. 3 for the general categories of devices.

Fig. 3.
figure 3

Types of smart home devices

Full size image

3.4 Qualitative Analyses

We conducted semi-structured interviews of 40 smart devices consumers to understand their perceptions of and experiences with smart home devices. Three NIST researchers initially coded a subset of four interviews and then met to develop and operationalize a codebook to identify concepts within the data. Based on the codebook, we performed iterative coding on the transcripts. We met and discussed the codes and progressed to the recognition of relationships among the codes as we examined patterns and categories to identify themes in our data.

We were interested in a subset of the questions in the in-depth interview to study whether the consumers in our study understood what IoT was and what made devices and homes smart. We focused this paper on the following research questions:

  • What do smart device consumers think the Internet of Things (IoT) is?

  • What do smart home consumers think make devices “smart”?

  • What do smart home consumers think it means to have a “smart” home?

To answer our research questions, we coded the transcripts using the terms that we developed in our code book using a widely accepted grounded theory qualitative data analysis method [11]. We used two codes for what is the Internet of Things: connected by internet or WIFI (i.e., wireless fidelity) and the second code of: don’t know. For smart devices, we coded responses with the following codes: learn/machine learning; programmable, connected, automated, and collects information and acts. For smart homes, we coded responses using the following codes: programmable, connected, controllable, home has smart devices, automated, makes life easier, and learns over time.

4 Results

We asked participants a series of questions about their experience of smart home devices in their homes. For this part of the study, we were interested in what participants thought about: what IoT is, what made a device smart, what made a home smart, and what the relationship is between the IoT and smart devices. Example participant quotes are provided below to illustrate the concept.

4.1 What Is the IoT

We asked participants what they thought the IoT was. We coded the responses into two major codes: connected devices by internet or WIFI and didn’t know. Most of the participants answered that it was a collection of WIFI-enabled or internet connected devices and a few saying that they were not sure or hadn’t heard of the IoT.

Twenty-nine of the 40 participants responses were that it was connected devices by internet or WIFI. Participants said it was basically devices that connected to the internet.

P10: My thought about what the internet of things is, basically a bunch of networked single maybe double purpose devices, micro appliances just connected to a network and they kind of work with each other.

P11: I consider the internet of things to be essentially devices that are connected to the internet and specifically in many cases, devices that may connect to cloud services, devices that produce data, or that data is aggregated and collected, and then potentially used to make decisions.

Nine participants had never heard of the IoT or didn’t know what it meant. Participants thought it was more of a technical term that they wouldn’t use day to day or had never heard of it.

P2: The Internet of Things. I haven’t really used that word too much to be honest. I feel like that’s more of a maybe more technology term for … I don’t think it’s a colloquial term that people use day-to-day.

P31: I don’t know that I’ve heard that term, actually, and I’m kind of someone that should, I feel. I would assume it’s just kind of a generalized term of how the internet affects our lives, but I don’t know that I’m remotely close to the answer.

4.2 What Participants Think Makes a Device Smart

We asked participants what they thought made a device smart. We coded the transcripts using the terms from previous research used to define smart devices: learn/machine learning; programmable, connected, automated, and collects information and acts. Most of the responses fell into three code categories: learn/machine learning, programmable, and connected.

Participants mentioned that to be smart the device needed to learn over time from the commands the owner issued, that there was some natural language processing, some type of machine learning, or that the device could think for itself. For example, P1 said: “I guess it would be something like a Nest that over time it figured it… it knows when you leave, so it can turn things down.” P7 talked about machine learning: “Maybe some machine learning involved so we can just evolve over time, they get smarter over time.” P14 talked about the device thinking for itself: “What makes a device smart? One would think if they can think for themselves or perform an action.” P8 provided an elaborate response:

“So, smart devices, in general, they … I would guess would you say they have a Learning component, but I guess they also share an Internet platform where they are able to be updated, they are able to be networked, and they think… I don’t want to say machine learning. I don’t want to go that far. Basically, they adapt, I’m going to use my Nest thermostat, for example. It adapts to your usage…. Then I think it uses an algorithm or machine learning, in some way, shape or form, to then better configure itself to you needs.”

Twelve participants responded by saying that the because the devices were connected (i.e., internet or WIFI) that they were smart. P11 answered this way, “These are devices that don’t fully function purely on their own but need some sort of network connectivity to basically report information or to make decisions.” P18 put it simply, “Your ability to interface on the internet.” P27 mentioned Bluetooth connectivity, “I think it’s that they can be used outside the home or they can be used without having to press a button. It’s more you can use it via Bluetooth or whatever.”

Eleven of the 40 participants mentioned that the device was programmable and that is what made it smart. P5 expressed it this way, “I guess they’re programmable that you can instruct them to do things according to your wishes…” P13 talked about not only can the devices be remotely controlled but they could also be programmed, “They can be remotely controlled, but also can be programmed to react to external stimuli on their own.” P19 said that he is what made them smart, “..it’s myself that makes them smart… I guess what makes them smart is the programmable decision points that are programmed into the interface.” P28 responded that you can program them remotely, “You can program them, you can schedule a smart device in your home. You can remotely program them when you’re away from home.”

A few participants mentioned automated, ease of interaction, or electronics/technology as the reason why devices are smart. P10 said, “A device that is smart is one that functions without any human interaction or can function based off of sensors or timing or some kind of automated schedule.” P35 mentioned ease of interaction, “The ease of interaction with them.” P31 said technology made them smart, “I think that devices are smart because they utilize technology to do things that were once low tech.”

4.3 What Participants Said Makes a Home Smart

We asked participants what they thought made a home smart. We coded the transcripts using terms that were used in previous research to define a smart home. Codes included:

Programmable, connected (internet, WIFI), ability to control devices, home with smart devices, and a home that makes life easier.

Approximately half of the participants thought that a smart home was one that was connected (i.e., internet or WIFI connected). As P11 explained it, “I look at a smart home as a home that has some sort of network, ether wired or most likely nowadays, wireless that is most likely internet connected…”. P28 talked about the home being all wired up. P28 described it this way, “It means it’s all wired up. It’s connected to the Internet. You have smart devices. Everything’s interconnected”.

About a fourth of participants said a home was smart because it had a number of smart devices. As P4 simply put it, “A home that has some of these devices, to me would be smart or smarter home.” P10 said a smart home needs a certain number of devices, “I don’t consider my home a smart home. It has a lot of smart things in it, but I don’t think it’s smart enough to call a smart home. So, I think you have to have a number of smart devices in your house to qualify for that.” P15 thought it would be a home with just one smart device, “Well again, I’d say that’s simply a home that has any smart device in it, would probably qualify as a smart home.” P35 wasn’t sure but added, “I don’t know. A smart device in your home.”

Another 25% said that a smart home was programmable, responds to our commands or is controllable. P1 talked about the device responding to a command, “You could say, what’s the weather? And it tells you.” P40 talked about controlling the home, “Actually to myself it means more control over the home itself, especially when I am not there.” P1 described a smart home as one that is programmable, “It might turn the lights off after you go to bed, or it’s set up at a schedule so that people know that you’re still at home even thought you might be away.”

Some participants said that a smart home simply makes life easier. P31 expressed it this way, “Well, for me, it’s just having those devices in my home that are meant to make life a little bit easier, but also maybe even make my home a little bit more secure.” P41 talked about quality of life, “A home that’s connected to electronic devices that would make life easier, quality of life easier.”

Only two participants mentioned anything about learning. P12 said, “A smart home would be it knows to lock the doors when you open up the door and then close it behind you, setting room temperatures for you when you’re inside of the house of outside of the house, setting your refrigerator at a certain temperature or freezer at a certain temperature.” P29 talked about a home that anticipates his needs, “I think it’s a combination of a home that works well for me and also one that anticipates my needs, but also kind of runs itself efficiently.”

5 Discussion

In this paper we describe a subset of data we collected in a semi-structured interview study on what the IoT is, makes a device and what makes a home smart. We wanted to understand if participants understood what made devices and homes smart.

As PwC [11] found previously, participants were largely unfamiliar with the term “internet of things”. Similarly, the majority of our 40 participants heard of the term, and only nine said that they had never heard the term or didn’t know what it meant.

The majority of the PwC participants (81%) were familiar with the concept of a smart home device and used terms like connected devices, makes life simpler, and collected and monitored data to increase efficiency. Possibly because our participants were all smart device consumers, they were a bit more experienced with smart devices. They described what made a device smart by talking about machine learning and natural language processing, connected devices, automation, and programmable.

Participants also used similar words to describe what made a smart home. About half of our participants said a smart home was one that was connected, about a fourth of participants said that it was a home with a number of smart devices, another fourth said it was smart because it could be programmed, and a few said a smart home made life easier. Only two participants mentioned that some learning component made a home smart.

Makridakis (2017) argues that within the next decade, the AI revolution will have an even greater impact than both the Industrial and Digital Revolutions [12]. Pessimists worry that as AI machines become smarter than humans are, that they will be making our decisions for us. Others are concerned about increased unemployment that may change drastically with the widespread introduction of AI technologies leading to massive job reductions [13]. Scientists like Etzioni (2016) do not believe that AI is a threat to humanity [14]. But there is little doubt that during the next decade that AI technologies will greatly impact our societies and lives in general.

In the next decade of smart homes, more and more AI will be driving these intelligent devices and homes that integrate these devices. No doubt, this could potentially impact the privacy and security of homeowners as well as trust they have in the devices when these devices begin making decisions for us and control not only homes but our lives as well. As Stephen Hawking has warned that the creation of powerful artificial intelligence will be “either the best, or worst thing, even to happen to humanity” [15].

6 Limitations

Our study may be limited in generalizability because of the limitations of in-depth interview studies (e.g., recall and social desirability biases). The participants were well-educated individuals living in a high-income metropolitan area and may not be representative of the U.S. smart home device consumer population. Our study also does not capture the perceptions of consumers who do not have or choose not to have smart devices in their homes. We realize that non-adopters may have different perceptions of AI and smart home devices. However, there is a lack of research in this area and even with these limitations, our exploratory study does shed some light on what smart home device consumers think the IoT is, what makes devices smart and what makes a home smart. Subsequent surveys of a broader population and specifically about these research questions would be of benefit.

7 Conclusion

Let’s imagine that it is nearer the end of this decade and AI has been introduced in most if not all smart homes and devices. Your intelligent home knows that you want to awake to new music on your music channel, so it wakes you up to a song on your list. But you never set an alarm because your intelligent home knows you have a spin class this morning because it checked out your workout goals. Your intelligent home checked the availability for a class at the local cycle gym, and when it found a suitable class, it scheduled you and paid for it as well. The home calculated your travel time and set your alarm appropriately the night before.

You get out of bed with your eyes barely open and you can smell the coffee brewing downstairs in the kitchen. Your refrigerator made sure that you had your favorite yogurt, and had it delivered earlier in the week. Your home knows that you have gained some extra weight, so it cuts down the portions and alerts your doctor that your cholesterol levels were elevated. You head to work after your spin class.

After a long day at work it’s time to return to your intelligent home. Your car alerts you home that you are on your way and your ETA is 34 min. Your intelligent home knows exactly what music to play and the temperature that you enjoy. You pull into the driveway and walk up to your door and your home scans your retinas and determines that indeed it is you. Your intelligent home decides that you should have salmon with steamed vegetables this evening to meet your health goals. After dinner you retire to your den to watch your favorite show. Oh, and by the way, you don’t have to clean up because your home robot will clean up after you go to bed.

Are you comfortable with your intelligent home recording your every move and making decisions for you? Currently there is little research exploring what consumers think is the Internet of Things, what makes a device smart and what makes a home smart, or concerns consumers have about their homes recording their homes, lives, and making decisions for them. More research is needed to understand how consumers perceive artificially intelligent homes and what concerns consumers may have with AI embedded devices.

Disclaimer

Certain commercial companies or products are identified in this paper to foster understanding. Such identification does not imply recommendations or endorsement by the National Institute of Standards and Technology, nor does it imply that the companies or products identified are necessarily the best available for the purpose.