Photo: @axel.bueckert via Twenty20
What makes voice assistants creepy?
MarketingCharts staff
Through a special arrangement, presented here for discussion is a summary of articles from MarketingCharts, which provides up-to-the-minute data and research to marketers.
Almost half (45 percent) of 5,000 consumers in Europe and North America surveyed by Selligent Marketing Cloud say they use voice assistants such as Amazon.com’s Alexa or Google Home. Yet with increased use, concerns are growing.
More than half (52 percent) find it creepy when brands target them with ads based on what they have recently asked their voice assistant. An even larger portion (69 percent) think it’s creepy when a brand targets them for ads based on what they have said in conversations without prompting the voice assistant device. In both cases, only minorities felt these were helpful rather than creepy targeting practices.
About half (51 percent) of respondents worry their voice assistants are listening to them without their consent. Younger adults (58 percent of 18-24-year-olds and 57 percent of 25-35-year-olds) tend to be more suspicious of voice assistants listening in uninvited. And although about half (48 percent) of Gen Xers (36-54-year-olds) have concerns about their voice assistants, eavesdropping only concerns slightly more than one-third (36 percent) of Boomers.
That being said, 80 percent of both Gen Zers and Millennials and 70 percent of Gen Xers say that a personalized experience is very important. Fifty-two percent of Boomers think so.
Respondents in all age groups feel that it’s helpful when brands do personalize in some ways, such as:
- Offering deals on items they have searched for but not purchased (64 percent);
- Proactively recommending products based on previous purchases (64 percent);
- Inquiring about how they liked their last purchase (71 percent);
- Targeting ads based on what they have recently liked/favorited/saved/retweeted on social media (53 percent).
RichRelevance’s 4th annual “Creepy or Cool” survey that came out in May also found concerns about voice assistants.
More respondents found it creepy (41 percent) than cool (32 percent) that voice assistants provide personalized product information and suggested products for them and their family. Related to voice assistants’ promise, significantly more respondents found it creepy (69 percent) than cool (14 percent) that companies understand their shopping habits so well that they’re able to use artificial intelligence to choose and automatically order products on their behalf.
- What Makes Smart Speakers and Voice Assistants Creepy? – Marketing Charts
- Selligent Global Connected Consumer Index – Selligent
- Consumers Remain Wary of Advanced Retail Personalization, Though Some Attitudes Are Evolving – MarketingCharts
- Creepy Or Cool 2018: Richrelevance Study Finds 80% Of Consumers Demand Artificial Intelligence (Ai) Transparency – RichRelevance
Discussion Questions
DISCUSSION QUESTIONS: Is there a way for voice commerce to overcome privacy concerns raised by consumers? What are the most useful ways to personalize without exceeding the creepiness threshold?
Poll
BrainTrust
Ken Lonyai
Consultant, Strategist, Tech Innovator, UX Evangelist
Mark Price
Adjunct Professor of AI and Analytics, University of St. Thomas
This is a classic trade-off between convenience and privacy. For voice commerce to be effective, it needs to monitor what people say. And while it is not substantially different than online privacy issues, there seems to be a much more personal, visceral response to the creepiness of voice commerce which seems to be related to the fact that voice commerce devices monitor people continuously – as opposed to online behavior. In order to reduce the creepiness factor of voice commerce, I suggest manufacturers consider clearer explanations about how the voice data is being collected and providing consumers with a clear way to shut off listening.
Agree!
Mark, I really resonated with your message. I completely agree. I feel that consumers should have total control over their information. Your insights are extremely pertinent.
The short answer is for these devices to only make suggestions when asked to. Of course, the issue is that all that money that comes from collecting users’ information will go away. I admit I can not understand the attraction to having a device in you home that can be listening to everything that you say whether addressing the device or not. Reminds me of the Miranda warning: ”Anything you say can and will be used …”
Consumers need to be able to physically turn the devices on/off so they can feel safer. Still, there is always the looming possibility that some malicious genius will figure out how to break into them. Voice assistants may be very helpful to physically-impaired consumers. The rest of the world does not need to get any lazier than it already is.
I find it interesting that those who worry most about privacy also want a more personalized experience – I would have expected an inverse correlation. The line between creepy and useful is constantly moving, and gets narrower with younger consumers – marketers need to respond with transparency and full disclosure.
Although I do not put anything past any of the companies offering intelligent assistants, based on known research and a deeper understanding of user profiling mechanisms, I’m extremely doubtful about surreptitious listening. Rather (and this would require an extensive response) users leave enough behavioral indicators directly and indirectly, especially through demographic matching and matching to associates, that it SEEMS like the assistants are spying.
That said, it would behoove these brands to finally shed light on their practices around privacy and data gathering/usage, to finally quell user concerns across all channels. By not doing so, they are hampering the potential of intelligent assistants and ultimately their own profitability. Additionally, they are flaunting with bringing more legislation.
So the short answer is spooky equals secrecy, delight equals transparency.
Simply put, voice assistants will ALWAYS be creepy. The question of whether this substantial privacy hurdle can be overcome is complicated but also simple. If enough value is created the majority of people will get over it. Credit cards have been breached and social media gives out our data, but we still are addicted to both. Value = stickiness. To fully get over this the model may need to change; perhaps the voice assistant companies should pay consumers to have the devices in-home versus the other way around.
There’s a fine line between awe and eww. Part of the “creepiness” of voice assistants lies in the “uncanny valley.” Masahiro Mori’s thesis is that as a robot or voice assistant becomes increasingly “human” observers’ emotional response to the voice becomes increasingly positive and empathetic, until it reaches a point of “too human” when the response quickly becomes strong distaste and “creepy.”
FYI – uncanny valley dates to 1970. It’s less relevant now, especially for those who have human-like technologies around them from early childhood.
Personalization is OK (as it is when customers are surfing online) but receiving ads based on what consumers have said in conversations without prompting the voice assistant device is a strong no-go. The question is also how to break up the oligopoly of Amazon, Google, Apple etc. and to open voice to everybody. It cannot stay like this.
Those are HUGE distrust numbers. Amazon and others should pay close attention. Unfortunately, they don’t seem to be.
I’ve never seen the value of voice commerce. Voice recognition remains so poor that even a simple attempt to dial by voice brings up wrong numbers quite often.
Which leads to my gut response when I read the numbers: Why do these people have a voice device at all? With distrust this high, seems they should step away from the brink and back to a cozy, private world.
It sounds to me like you’re arguing against yourself, Doug. People have the devices and are trying them for voice commerce because they want voice commerce. Have you ever seen an e-commerce web site from 1996 and compared it to one from today? It’s early days for voice commerce and it too will improve drastically, especially when it finally evolves to be part of a multimodal commerce interface/experience.
Lastly, you are incorrect that “Voice recognition remains so poor.” The best systems (including those from Apple, Google, and Amazon) have voice recognition capabilities that surpass human voice recognition. They can be hampered by environmental noise and speaker distance, but that is being iterated continually.
Perhaps. We won’t really know. But my wife still finds that voice recognition can’t successfully recognize when she asks to call me — and that’s after 10 years of intense development. (Which fits with the Amazon mythology. Few people I know do anything but laugh at Amazon’s attempt to recommend books to them … despite extensive data.)
Fundamentally, I’ve never had the faintest urge to want to order something by voice. I’m sure there’s an extreme fringe who do. But my bet is on the core center of shoppers and buyers — who still don’t quite understand why they’d want it or why they’d trust it to do it right.
Perhaps the Bezos meme (that he accidentally bought Whole Foods when he tried to order whole fruits) dominants perceptions. 🙂 )
Cheers.
It makes me giggle when the masses flock to the latest tech gadget that is supposed to read their minds… and then when it does, they get offended and paranoid that it knows everything about them. For me, I still get up off the couch to hit the light switch. Google, Apple and Amazon don’t know that.
Creepy? That is when I am not talking to Alexa but she thinks I am, so she gives me an answer! That reinforces that, like a pet, she is listening to every word and sifting through, but unlike a pet she is recording it all. There was a story about a family watching a news story on TV about a child buying a dollhouse with Alexa. The Alexa of the family who was watching heard the story on TV and bought a dollhouse!
I love the ease of Alexa, calling out a task and having her address it. She has not yet recommended items we might like to try but I do know that when she is turned on she is listening and cataloging information about my family. That’s a trade off we have to make if we want to use her services.
The creepy part is when she randomly talks or plays music with out being asked. That’s always a good time.
It’s a funny thing because it takes a while to realize that if the harmless looking device can listen and then speak to you, that information doesn’t just sit there. And now, with the current movement towards more privacy scrutiny on the social platforms, you can’t help but thinking, “well, who does have all that information I just gave them?” Right. Who does have it? Having said that, given consumers’ rather lackadaisical attitudes towards privacy and the ease of voice shopping, I still think it’ll win out in the end. It’s just too damned easy. Maybe though, we’ll start to read those agreements now.
And for what it’s worth, see the movie Anon. Terrific thriller along the lines of the Black Mirror stuff.
Seriously, suspending disbelief that Alexa is always listening is to ignore the facts. Alexa targeted advertising is the tip of the iceberg, as they say. Whoever they are, they seem to be right. But creepiness is not confined to Alexa. My iPhone asked me out of the blue the other day how I liked lunch at a restaurant I had visited the day before. My iPhone, without being asked, guesses where I am going and how long it will take to get there. Who or what else is observing our day-to-day life? Likely all this data will become the basis for AI enabled “listening” to every aspect of our lives. To what end?
Asking a “voice assistant” for something is akin to a Google search. Teach the customer that, similar to a Google search, they will receive targeted ads. If that “creeps them out,” then the customer has a choice. However, browsers like Chrome give the customer an option of going incognito, which gives them the privacy they crave. Maybe the voice assistants can offer the same. That said, there are great advantages to letting the VA learn about your habits and usage patterns. It allows for a more personalized experience. So, the customer has a choice, just like they have a choice to use (or not use) Google, Facebook, etc. The companies putting out the VA (Amazon, Google, etc.) need to educate their customers and give them that choice.
Privacy is a value exchange. It wasn’t long ago that posting family pictures and life events on Facebook was insane; today it’s the status quo. The value outweighed the privacy cost. The same thing will happen here. As voice assistants become more useful (personal) and integrated in our lives, we’ll start to accept all the ways we surrender our privacy to get it.
Ken is right that transparency about how voice devices work may alleviate some of the creep factor. But it may add other discomfort. The makers probably reason that they don’t want to emphasize the extent to which their sensors are listening and gathering user info by being too forthcoming.
Unsolicited voice communications from these devices could have the effect of highlighting their intrusiveness. When visitors to my home hear my Nest smoke/CO detectors announce a self-test they are visibly startled. If an Echo device chirped that it may be time to re-order toilet paper, I might be tempted to hit it with a hammer.
Just as we humans have had to learn a thing or three about email etiquette and texting etiquette, I suspect the designers of voice assistants will need to establish some etiquette guidelines of their own. Perhaps the AI will supply these over time.
I propose that Alexa, Siri, and Google add new “skills” that allow a human user to say “I don’t like what you just did,” or words to that effect. Nothing like real-time response data to hasten the learning cycle.
Offline devices that interconnect through a closed loop system with easy to master security will gain acceptance and growth much faster than what we have now. The collection of consumer information with or without consent is repugnant to the market place. Companies ignore this and insist it is to better serve. The consumer now knows that this is a lie and consumer information is bought and sold all day every day. This outpour of private lifestyle information may bring back cash and Brick & Mortar stores in the not too distant future.
This is definitely a “slippery slope” indeed. I am optimistic about the future applications for voice technology however, there is a massive consumer tradeoff of convenience for privacy. On one side of the fence is the amount of ubiquitous information readily available at your fingertips. Conversely, consumers are sharing personal and sensitive information with these devices that companies have already admitted to listening to under the guise of improving performance. I am looking forward to seeing consumers having more control over sharing their information.
The inherent fear of individuals who don’t have clarity into what’s being done with their data is a scary proposition. Would you as a consumer ignore someone looking over your shoulder and eavesdropping on you 24/7? Studies show people who believe they are being watched or listened to have much higher levels of stress.
I believe voice commerce will struggle. The conveniences don’t make sense for retail except for replenishments scenarios. The expectation the machine will know what your preferences are or can cater to your preferences usually requires some kind of sensory factors, such as visuals or descriptions. Enabling this through voice eliminates the convenience factor. Add to that the “creepiness” of 3rd party organizations having full transcripts of your conversations berating your kids or sharing bank account or health information and digitizing things that have been traditionally highly personal, the scenario becomes even less compelling in terms of real convenience.
At this time, voice commerce doesn’t provide adequate conveniences and doesn’t alleviate fears of highly personal information security access. Until both of these take place, we won’t see it in the mainstream.
It’s clear that we are in the middle of a sea change in the way consumers relate to technology. First the Millennials and now the Gen X’ers are becoming more and more accepting of the leverage of their data to improve their experience, either by anticipating their needs or suggesting complementary items, discounts, etc. As these two groups form more and more of the collective buying power, the role of anticipatory technology continues to increase.
At the same time, research suggests that consumers consider the monitoring of their conversations to be a “red line” that retailers and e-commerce merchants should be wary of crossing. At least for the moment, consumers in the younger segments who are mobile-fixated tend to be comfortable with the leverage of data to improve their experience when that data is voluntarily provided. And the definition of voluntarily provided seems to be shifting over time.
In the next five years, I anticipate that even conversations will be considered fair game — just that more patience is required.
The issue, as always, is consent.
Are consumers able to designate when they give consent to be heard, and when they do not? Even if the voice assistants claimed to deaf until summoned, would consumers believe that?
Social media companies and web data collectors have abused private information in the past, why not again?
It seems naive to trust a 25/7 listening device, even if it can recommend a few products.