Building Trust in Artificial Intelligence
Jan 5, 2017
Lately, one of the most fervent conversations in the field of artificial intelligence has been about trust. As intelligent agents have moved out of labs belonging to researchers and business analysts and into the lives of everyday consumers, discussion about the limits, biases, and safety of AI tech have evolved from sci-fi theory to a real and serious debate. Given the fact that the outcomes of this discussion (i.e. investment, regulation, and cultural paradigms) will impact the bottom lines of businesses driving AI technology forward, it’s important that they play a role in the discussion—positioning their offerings as trustworthy, first and foremost.
Earlier this year, Uber took charge of this conversation in anticipation of the debut of their self-driving service. According to a 2016 study by the University of Michigan, just under 16% of Americans reported that they would want to ride in a self-driving vehicle. To overcome riders’ barriers to adoption, Uber deployed a communications platform aimed at fostering trust in their technology.
Uber’s campaign began with a media event in which rider safety was a key theme. In early 2016, they announced a partnership with Volvo to produce a self-driving XC90 SUV Ubers. This would help them borrow consumer confidence from Volvo brand’s enduring legacy in safety (and the car’s 5-star crash rating from the National Highway Traffic Safety Administration).
When Uber’s driverless vehicles first hit the streets in September, the company released a video. In it, an autonomous car navigates Pittsburgh, safely taking a right-hand turn in front of three old ladies crossing the street and yielding to a pedestrian. The campaign for trust extended to the ride experience itself. Passengers were given a tablet that displayed what the car saw in real-time. And the human in the driver’s seat was as much about making riders feel safe as it was about ensuring their safety.
By all accounts, Uber’s autonomous vehicle pilot in Pittsburgh has been a success. Only a couple incidents occurred, and both were attributed to human error (that is, until one of Uber’s autonomous cars ran a red light in San Francisco this month). Most importantly, Uber was able to persuade riders to entrust their safety to an autonomous car.
But, for Uber CEO Travis Kalanick, the Pittsburgh debut was just the first of many milestones in the journey to becoming “a trusted robotics company.”
“There’s going to be a long transition,” he said in a February 2016 interview at TED Vancouver. “This is a world that’s going to exist, and for good reason. A million people die a year in cars. And we already looked at the billions or even trillions of hours worldwide that people are spending sitting in them, driving frustrated, anxious. And think about the quality of life that improves when you give people their time back and it’s not so anxiety-ridden. So I think there’s a lot of good.”
You could say Uber’s marketing challenge wasn’t entirely new. Safety has long been a pillar of brands in the automotive and transportation verticals, and consumers are used to being a passenger, deferring control to either humans or machines.
But, it’s in this “deferring of control” where the unique marketing challenge lies. In Pittsburgh, Uber persuaded riders to let artificial intelligence be their chauffeurs on the open road—where human drivers often experience incident, injury, or even death. This is the universal challenge of artificial intelligence: marketers must persuade buyers to defer sophisticated decision-making to a machine. They have to convince people that AI can do it – or even do it better.
To do this, marketers need to address the buyer’s barriers to adoption.
Psychologically, the buyer must trust the technology. To trust it, buyers must understand it. To understand it, buyers must be able to see it. To see it, they must be able to relate it to what they already know.
Sociologically, the buyer must understand how their use of AI impacts the people around them. They need to feel that they’re behaving within cultural norms. Logistically, the buyer has to understand how the technology fits in their lives, as well as how to interact with it.
Additionally, marketers must be transparent in that AI technology will fail from time-to-time. Prime directives around safety and ethics are essential in ensuring those failures happen within “well-founded boundaries.” Marketers must ensure consumers are aware of those boundaries, as well as the rate at which they can expect to experience the technology’s failure.
Uber demonstrated that autonomous vehicles could function as expected. By showing riders what the vehicle saw, via a tablet, Uber helped them understand how the car navigated. It gave passengers visibility into how the car considered the people around it. That was about as sci-fi as the experience got. Otherwise, it was your standard Uber ride.
The world’s leading ride-sharing service is just one of many companies who are betting on artificial intelligence.
As Oren Etzioni, CEO of the not-for-profit Allen Institute for Artificial Intelligence, mused in a November article, Google, Facebook, and Microsoft Are Remaking Themselves Around AI, “This isn’t just window dressing.” Artificial intelligence is already core to many of the technology offerings 21st-century consumers rely on day-to-day.
If the marketers of artificial intelligence fail to establish trust in the technology and help consumers understand the boundaries in which it will operate (and address failures), we may someday find ourselves collectively experiencing what psychologists call a “breakage.” A deficit of trust and unanticipated calamity could engender a malaise or even wholesale rejection of artificially intelligent technology. In such an event, with all the economic, environmental, and scientific benefits we have yet to realize in AI, humanity would be the biggest losers.
As Harvard Business Review proclaimed in their November 29th piece “What It Will Take For Us to Trust AI,”“… building trust is essential to the adoption of artificial intelligence. And we believe that its adoption is essential to humanity.”
As a marketer of intelligent technology, your buyer’s journey to adoption may be obstructed by:
- a deficit of trust,
- a misunderstanding of how your technology impacts others, or how it’s perceived by given social groups (or society at large),
- or a lack of understanding as to how the technology fits in the buyer’s life and makes it better.
To overcome these barriers, you’ll need to communicate with buyers to help them overcome these barriers. To determine what messages you need to deliver, take the perspective of your customer and consider the foundational steps to adoption:
|Layer—What to Ask||Example: Uber|
|Existing Knowledge—How does your offering relate to other, less “futuristic stuff” in the buyer’s life?||Existing Knowledge—A self-driving Uber ride is just like a regular ride, but autonomous.|
|New Observations—How can you give the buyer visibility into how the intelligent system produces its outputs/behavior and if it’s behaving as expected, all in an accessible and relatable way?||New Observations—Uber gives riders visibility into the car’s light radar—seeing what the car “sees.”|
|New Understanding—How do you help the buyer understand the platform (what they’re seeing and interacting with)? Why is it coming to its outputs/behavior?||New Understanding—Uber gives riders access to a FAQ that explains the autonomous vehicle in way they can get.|
|Social Fit—How will the buyer’s use of the technology impact the people around them? How does their use match social norms?||Social Fit—Uber conducted a robust community and rider communications strategy that helped them understand the scope of the Pittsburgh pilot project.|
|Logistical Fit—How does the technology fit in the buyer’s life or work? How do they use it? Why is it better than alternatives that don’t use advanced technology?||Logistical Fit—Uber delivered real rides.|
Once you’ve identified what your buyer wants to know before adopting your offering, determine the messages, content, and experiences that will help them learn. Then, deliver it.
As innovations in AI technology quickly evolve, so too will the understanding of its impact on people. To keep up with the latest issues around trust and AI, technology marketers should follow a few organizations leading the discourse:
- Stanford University has chartered what they’re calling the “One Hundred Year Study on Artificial Intelligence” (or ‘AI100’). Researchers from top universities across the country have joined the initiative to “study and anticipate how the effects of artificial intelligence will ripple through every aspect of how people work, live and play.”
- In September, tech giants Amazon, Google, Facebook, IBM and Microsoft formed the “Partnership on AI,” which is bringing together academics, non-profits, and specialists in policy and ethics to inform their own work, as well as innovate related public policy. Similarly, Carnegie Mellon, recently received a $10 million gift dedicated to the study of ethical issues posed by AI.
As marketers, let’s overcome buyer’s barriers to adoption, persuading them to choose a life bettered by advanced technology.