The Key Skill Robots Will Need To Master Before Taking Your Job

Before they can take over bigger chunks of the workforce, robots may need to learn to make small talk. Few of us seem willing to let them.

The Key Skill Robots Will Need To Master Before Taking Your Job
A Kokoro Company Ltd. humanoid robot stands at the reception desk of the Henn na Hotel, operated by Huis Ten Bosch Co., a unit of H.I.S. Co., during a media preview in Sasebo, Nagasaki Prefecture, Japan, on Wednesday, July 15, 2015. [Photo: Akio Kon/Bloomberg via Getty Images; App Photo: Buena Vista Pictures ("Bicentennial Man", 1999)]

The risk that your job will be automated out of existence depends, of course, on the job you do. For many, that’s already happened–typically in roles and industries where the name of the game is eliminating human error and improving efficiency.


But in order for artificial intelligence to take a much bigger bite out of the knowledge-economy workforce, the technology may need to start behaving more like humans, not less. And that will mean mastering one key behavior: small talk.

Talk We Want And Talk We Don’t

Sociolinguists involved in the Language in the Workplace Project at Victoria University of Wellington in New Zealand have discovered that people switch naturally between “transactional” talk–such as discussing a business goal–and “interactional” talk, like when you encourage or show concern for a distressed coworker. We do this toggling back and forth with our colleagues all the time without even realizing it. It’s at the heart of how we communicate.

So far, this kind of mental flexibility isn’t something machines–designed to execute one type of task, consistently and flawlessly–can currently manage. And they won’t, according to machine-learning expert Geoffrey Hinton, until machines’ neural networks of 1 billion synapses come closer to the 1,000 trillion synapses of the human brain.

For the moment, anyway, the technology that’s best equipped to supplant human workers is still at a “rudimentary stage,” says Robert Stephens, cofounder of, a leader in the business messaging and bots space, whose clients include Fandango, 1-800-Flowers, and Hyatt. When it comes to group chat, he points out, machines can’t yet distinguish between humans talking to each other and people instructing a bot directly.

Stephens sees the workplace value of bots as “more anticipation than automation,” helping eliminate some of the decision-making from humans’ most routine tasks. And it’s these sorts of things that Google for Work, for instance, seems most preoccupied with, foreseeing a workplace of the not-so-distant-future in which our smartphones act as personal assistants, automatically reorganizing our schedules, dropping us reminders, calling us cabs, and serving up documents in a trice.

But no matter how sophisticated they become in these roles, Stephens tells me, many of us simply “don’t want chattiness” from a machine. Having small talk with our smartphones (or whatever technology come to replace them) will only slow us down. The tech companies like Google and others that are building AI tools for the workplace seem to understand that. After all, this belief is already deeply ingrained through our interactions outside the workplace. In her book Small Talk, the University of Cardiff’s Justine Coupland found that banking customers preferred machines to handle their requests quickly and efficiently without pretending they’re anything like us. Their value was in being transactional and not interactional.


“Naturalness is not an absolute but an emergent property of dialogic context,” Coupland writes–in other words, we can do without it in many situations, and often prefer to. Coupland has found in her research that while callers speaking with human agents willingly engage in small talk–creating a sense of friendliness that can actually enhance their perceptions of the company–with a machine it feels to many of us “at best an irrelevance, at worst an impediment, to full usability.”

Plus, some AI development might be driven by other considerations altogether. As The Verge‘s Casey Newton wrote after interviewing Microsoft CEO Satya Nadella about the company’s intelligent apps and services, “For now the decision around bots and AI remains driven by the industry’s desire for a profitable new platform, rather than consumer demand for the services they provide.”

Why The Human Element Still Matters

At the same time that demand for efficient, transactional, anticipation-driven workplace tools is rising, so is the need for the skills we seemingly don’t want robots to perform.

Recruiters and employers say they’re looking more arduously than ever for candidates with great “soft skills”–and worrying that too many new grads don’t have them. Emotional intelligence is said to be among the fastest-growing job skills, and some experts say the ability to collaborate and listen thoughtfully can even protect your position and help advance your career over the next decade as automation continues apace.

It’s no surprise why. Small talk–even gossip, up to a point–is well-known by psychologists to foster community and inspire trust. This “interactional” chat is an essential building block of a functional work culture and crucial to employee productivity–at least as long as “employee” means “human.” So before robots can remove people completely from entire industries, they’ll need to learn to work a lot more closely with us–and, subsequently, behave a lot more like us–which technologists and ordinary people alike seem pretty disinclined to let them do.

Even if that were to change, the obstacles to a more humanlike AI experience in the workplace are steep and numerous. As Sharon Schweitzer and I explain in our book Access to Asia, the way we think, feel, and behave is far from universal. Take Japan, for instance, a leader not just in robotics and AI but in humanoid robots in particular. Few other countries have as enthusiastically embraced robot receptionists, babysitters, and elderly companions.


But much of that success may be thanks to demographic and cultural circumstances and won’t be so easily replicated elsewhere–least of all in the workplace. For one thing, Japan has the highest percentage of people over 60 and a relatively tight immigration policy, making the demand for technology to supplement a strained nursing and caregiving workforce especially high. There also may be the influence of Shinto philosophy to consider; as one University of Michigan anthropologist explains, “In contemporary Euro-American culture . . . what is living has spirit or soul; what is not living does not. Such distinctions are not so clearly made in the Japanese way of thinking.”

The point is that we may be wrong to see the threat of automation as a global inevitability when cultural and other quirks make it anything but. As Ken Anderson, an ethnographer at Intel, told The Atlantic in 2013, “Technology companies as a whole are in danger of being more disconnected from their customers than other companies. Our mind-set is that people are just like us, and they’re really not.”

Replacing large swaths of the knowledge economy with machines will require grappling with these issues, not blowing past them. And to do that effectively, our future robot-colleagues will need to learn how to chat with us, much the way we already do with each other. That prospect seems pretty far off, despite the onslaught of innovations we’ve now grown used to seeing pile up around us. Automation is sure to continue, but chances are it won’t take the forms we expect–often, as in the case with small talk, for reasons we hardly pause to consider.

Liz Alexander is a futurist and cofounder of Leading Thought, which passionately helps prepare human beings (only) for remarkable futures and collaborates with companies that want to be future-smart. Connect with Liz on LinkedIn or follow her on Twitter at @DrLizAlexander.