APEX Insight: Robots, like machines, were designed to perform repetitive tasks, but as they evolve, so does their understanding of humans – and our understanding of them.
We tend to think of machines and humans as opposites, unlikely bedfellows at best. One: hard, plastic and unfeeling. The other: soft, supple and empathetic. The last of these qualities is at the core of what many of us think it means to be human – we care.
But feelings aren’t just signs of human depth; they’re tools, mental shortcuts, that inform and guide our behavior, explains Christopher Noessel, global design practice lead, Travel and Transportation, IBM. “Disgust, for example, saves us from having to perform a scientific test on whether a given thing is toxic. It’s faster when we can just ‘go with our gut,’ and steer clear,” he says. “Another emotion, fear, helps us put aside other concerns to prioritize our physical well-being.”
“We need to find a way for robots to feel when they’re in one of those gray areas, and quickly, gracefully exit and pass over control to the right human being.” — Christopher Noessel, IBM
Artificial intelligence (AI), on the other hand, simply doesn’t need emotional cues to do many of the things that humans do. “A robot can access the entirety of toxicology at a moment’s notice to assess a substance’s toxicity, or continually run predictive algorithms to stay ahead of threats,” Noessel says. But even with big data, robots aren’t yet able to exercise empathy and handle ethical situations as a human would. “We need to find a way for robots to feel when they’re in one of those gray areas, and quickly, gracefully exit and pass over control to the right human being,” he says.
A LEARNING PROCESS
Until machines can intuitively perform human intellectual tasks, or at least know when to pass them on, they’ll likely continue to perform more repetitive tasks, like carrying baggage, processing check-ins and navigating travelers through sinuous airport terminals. Kevin Kajitani, an intrapreneur at the Digital Design Lab of All Nippon Airways (ANA), says the Japanese airline is deploying Pepper, SoftBank’s emotionally intelligent humanoid robot, in areas in airports where travelers often make wrong turns – a task previously conducted by a human. “All day, there was someone just standing there, saying, ‘Go this way if you have a connecting flight.’ That’s not something a human does very well; they won’t be satisfied with their job,” he says. “But a robot can do that all day long and be completely ‘happy.’”
One argument for the restriction of AI to production-heavy and information-processing jobs is that robots lack the ability to adequately perceive social, cultural and emotional nuances. But this isn’t completely true, says Junichi Shimizu, project leader of Haneda Airports’ robotization project. AI can be programmed to read subtle cultural cues such as the difference between a bow signifying “thank you” and one meaning “I’m sorry,” Kajitani explains. “The facial expressions, muscle movements, and speed and angle of the bow can be learned by AI and other robots.”
Noessel agrees, explaining that a cross-culture video corpus of expressions and their meanings could plausibly be built for robots to assess cultural norms, predict emotions and adjust behavior accordingly. “I think IBM Watson APIs [application programming interfaces] will be used to create such a system,” he says. “We are already able to use those APIs to detect emotions in written text or spoken language and either modify a system’s response or alert a customer service representative about it.”
Machines may, as of yet, be unable to solve issues that rank high on the emotional spectrum, but they can nonetheless help customer service agents care “better”– if given the opportunity. In addition to taking on airport functions that can be automated, AI – when seen as a way of extending human capabilities rather than replacing them – can empower humans with a deeper understanding of the passenger experience and the means to improve it.
THE JAPANESE WAY
This type of more collaborative human-robot relationship model is firmly in place in Japan, where robots have long been perceived as human aids, not opposites. There are several reasons why Japanese culture might embrace robots more than American culture, says Kate Darling, self-described “Mistress of Machines” and a research specialist at the MIT Media Lab. “Animism doesn’t distinguish between living and non-living the way that Western religions do, so the belief that objects can have souls may make it easier to accept robots and, in particular, robots created in the image of something lifelike.”
Indeed, more-humanoid airport robots like recently upgraded EMIEW3 – now equipped with active-learning dialogue-data-based AI technology – and Pepper hail from Japan-based Hitachi and SoftBank, respectively. And San Jose International Airport’s dancing humanoid robots, named Norma, Amelia and Piper, were the brainchildren of Future Robot, a robotics firm based in South Korea. Animistic beliefs are more prevalent in South Korea than they are in the US, where seeing an inanimate object imbued with humanness can be unsettling.
“We have to decide culturally if we want things that don’t have emotions to exhibit fake ones,” Noessel says. “I’d want to test that directly with travelers, but my gut says that it can feel creepy and upsetting when the robot is getting something wrong.”
In Japanese culture, robotics has long been tied to the field of psychology, perhaps making it easier for people to be more accepting of machines that perform human intellect and emotion. “For example, Professor Hiroshi Ishiguro talks about how to get people on the receiving end to engage with the robot. It’s very deep into human psychology and interaction user interface,” Kajitani says. Kajitani’s colleague, Akira Fukabori, intrapreneur, Market Communication, ANA, agrees, saying this focus on the human-robot relationship is “really Japanese.”
“Astro Boy is a popular reference in Japan and is a positive story about a robot, whereas Western science fiction has tended more toward dystopian stories of robots taking over.” — Kate Darling, MIT Media Lab
Darling adds that the representation of robots in Japanese science fiction and pop culture may be another reason why the country is more amenable to the presence of android helpers. “Astro Boy is a popular reference in Japan and is a positive story about a robot, whereas Western science fiction has tended more toward dystopian stories of robots taking over,” she says.
According to Shimizu, this dichotomy also plays out in the physical attributes of robots being used in Japan versus those in the US. Talking about Caiba, one of the concierge robots currently being deployed in Tokyo’s Haneda Airport, he says, “It’s compact and cute, and has cute little features … It’s more like a stuffed animal, instead of the AI in the US, which are more serious, more robotic, harder and colder.”
Cultural inculcation and cute design aside, nothing beats the stimulus that comes from sheer necessity. Japan’s population of 127 million is forecast to shrink by one-third in the next 50 years, with the number of over-64-year-olds – currently estimated at 25 percent – expected to reach 38 percent, according to a recent government report. Kajitani explains that as a result, the younger working population is going to have to output twice as much as the current generation. “I don’t think the technology is seen as so scary in Japan because it is so necessary in that culture right now. We just don’t have enough people.”
An analog to Japan’s labor market decline can be seen, albeit at a smaller scale, in the aviation industry’s impending worldwide pilot shortage: Boeing predicts the industry will require 637,000 pilots in the next 20 years, with industry growth projected at 4.8 percent per year. As a solution, the aerospace manufacturer has proposed a pilotless jetliner, which it says will be tested in a simulator by the end of this year and in a flying test bed as early as next. According to Boeing, the project will expand upon existing autonomous flight management systems by tackling challenges related to taxiing, incorporating aircraft into airspace and mitigating unplanned events – tasks that pilots usually intervene in.
“I don’t think the technology is seen as so scary in Japan because it is so necessary in that culture right now. We just don’t have enough people.” — Kevin Kajitani, All Nippon Airways
Though the technology is there, travelers aren’t necessarily ready to embrace it. A recent study by Swiss investment bank UBS found that while pilotless airplanes would save airlines an estimated $35 billion a year and lead to significantly reduced airfares for travelers, 54 percent of the 8,000 people surveyed said they would refuse to travel in a pilotless aircraft. “This lack of trust,” Noessel says, “is one of the reasons not to eliminate humans from systems.”
One of the main reasons we trust complete strangers with our lives more than we do equally proficient AI is we feel that we can relate, Noessel says: We trust that they operate with a similar moral compass and according to similar principles of self-preservation. “I trust the pilot at the front of the plane because I trust that pilot wants to get herself safe and sound to our destination, too,” Noessel says.
But trust can be built over time. “One of the ways we build trust in people is by giving them low-stakes tasks, seeing them perform them correctly, and seeing them recover from mistakes many times,” Noessel says. “So, some of the trust issues will be resolved generationally, as people experience robots successfully navigating higher-and-higher-stake tasks.” Having won human trust on repetitive activities, machines capable of detecting facial expressions, emotional states and other sociocultural nuances are entering the scene, thereby reducing the distinction between help from a robotic hand and that ever-elusive human touch.
“Programmed to Care” was originally published in the 7.5 December/January issue of APEX Experience magazine.