All the sessions from Transform 2021 are available on-demand now. Watch now.
With the current breakneck pace of innovation, it may seem like technology is on an aggressive mission to solve all of humanity’s most pressing problems.
And in some respects, we’re making great progress. We’ve broken tremendous ground in such areas as renewable energy, disease prevention, and disaster recovery. But when it comes to addressing important human-centric challenges — things like workforce diversity, unconscious bias, and employee and customer satisfaction — technology has a history of coming up short.
That’s because technical problems like jet propulsion or GPS are largely math- and physics-related, which is where computers (and programmers) excel. But solving human problems like employee engagement usually requires empathy, and that’s notoriously hard to codify. Humans are emotional creatures, especially when it comes to making decisions. First we feel, then we apply logic to help justify our emotional response, and, finally, we act. Thus, any attempt to help people make better decisions that doesn’t account for emotional factors is almost always destined to fail.
However, with recent advances in artificial intelligence, and especially natural language processing (NLP), we finally have the technological tools for tapping into the power and complexity of human feelings. This approach has major implications for how we design systems, and it’s leading to solutions with more humanistic points of view.
Programming for difference
Language is incredibly complex. From one person to the next, the slightest difference in someone’s experience or environment can shape how they express themselves. Dialects, genders, locations, and even seasons can change the words we use to convey an idea.
People are very good at accounting for these subtle differences. However, for computers, it’s a massive challenge. To come anywhere close to achieving human-level understanding, they need an enormous and rich set of language training data that spans countless examples of each variation in demographics, experiences, and backgrounds.
To see how this works in real life, just think about a teenager in California using the word “lit” (meaning “exciting”) when reviewing a new smartphone, and what that same word might mean in a review coming from a senior citizen in Massachusetts (perhaps “screen brightness”).
Reading between the lines
For the first time, we’re able to teach computers to understand not only the basics of what people are saying — by doing things like counting words or looking for specific phrases — but to intelligently “read between the lines” and get to the true intent and meaning behind our words. This, of course, is an important skill people have acquired over time as a function of empathy.
The common “satisfaction survey” is a classic example of technology’s limitations in solving even the most basic questions of how someone feels. In principle, it’s a powerful way to gain an understanding of how people feel about a product or service. In practice, it’s clunky, inaccurate, and long overdue for a remake.
Consider the survey prompt found on most store receipts: “Please rate your experience from 1 to 10 and share why.” Compare this with how a person would approach the same challenge — by simply asking “What’d you think of the experience?” and then inferring the “score” from the language used and the overall context. While people don’t need to ask for explicit ratings, machines do.
Looking in the mirror
In addition to helping us better understand one another, NLP can also give us a better understanding of ourselves. Language is the most detailed window into our thoughts and feelings. When technology can begin to understand us as we are (not as how it wants us to be), it can become a true partner to help us discover how best to grow and improve.
Take the dreaded performance review and the various biases that plague it. When you ask people in a work environment whether they might be biased, even subconsciously, they will vigorously deny it. However, studies of performance reviews show widespread, unconscious bias.
Analysis from my team showed that when men reviewed other men, they overwhelmingly used passive language (“they could be more proactive”). When these same men reviewed women, though, they often used finger-pointing language (“you should work on your attention to detail”). Using data-driven technology, we were able to provide further insight into this hidden bias that many of us unknowingly carry. Fortunately, AI can put us on the path toward correcting such biases by bringing them to our attention.
To solve the world’s most challenging “people problems,” whether through developing better products or enabling better understanding and more fairness in the workplace, we need technology to demonstrate empathy. When we leverage advances that combine both heart and mind, we can further develop and deliver the people-centric solutions we all deserve.
Armen Berjikly serves as senior director of strategy at Ultimate Software, where his expertise in human-computer interaction drives Ultimate’s transformative artificial intelligence platform and direction.
VentureBeatVentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative technology and transact. Our site delivers essential information on data technologies and strategies to guide you as you lead your organizations. We invite you to become a member of our community, to access:
- up-to-date information on the subjects of interest to you
- our newsletters
- gated thought-leader content and discounted access to our prized events, such as Transform 2021: Learn More
- networking features, and more