Subscribe / Unsubscribe Enewsletters | Login | Register

Pencil Banner

Meet ADP's 'business anthropologist,' putting human thought behind chatbots

Mary K. Pratt | May 30, 2017
Martha Bird's specialty is designing technologies that work for people, creating conversational user interfaces and chatbots that can serve different professional audiences across different geographies and different cultures.

How do you build for a diverse audience? I'm a big proponent of spending time with the people we're designing for and asking ourselves, "What would we want, how would we use this, how would I want to be treated?" A lot comes from that, essentially trying to understand what's relevant to others.

Can this thought process be taught? It can be learned. It takes a commitment to a certain kind of vulnerability of remaining silent, so you can actually listen. It's also about the way one listens: listening disengaged from self-interest. It's about stepping back and listening with respect. I call it "active listening."

Do enough IT people do this? I think on balance it's not one of their strongest aptitudes. I think tech folks are focused on writing code and developing algorithms but would do well to be more exposed to the end users and see the psycho/emotional/utility end. I find developers are amenable to that, but they don't get enough opportunity to do it.

You've also used the term "deeply hanging out" to describe your work. What's that? It goes back to the idea of respectful active listening. I'm not entering people's spaces with a clipboard and a camera, and recording like they were human specimens. I'm trying to understand what their expectations are in life generally and how does that actually relate to their work life.

I spend time in places where people make meaning with the tools they use. If it's a software system for practitioners, I spend time in their office or cube or wherever it is they're working and pay attention.

What are "contextually sensitive conversational cues"? In conversational interfaces, it's incredibly important to understand both your audience and your own voice, and how the tone will vary for a positive situation to a negative situation.

For example, the tone will vary for positive situation (like presenting a job offer) to a potentially negative situation (starting employment separation). If you don't adjust for these different audiences and scenarios, conversations can easily come across as unnatural or, worse, offensive.

What's a "consistent personality" in chatbots? Before we started creating conversations, we identified the key personality attributes we wanted for our conversational user interface. They include not only who was this system and what it was like but also what it wasn't.

It was important to be proactive, but never pushy, and that being friendly was authentic but was not forced. We refer to those attributes regularly while we create new conversations and content. We can ask ourselves, "Is this something the system would say?" If the answer is no, we could adjust the conversation. That's what it's like to ensure consistent personality.


Previous Page  1  2  3 

Sign up for CIO Asia eNewsletters.