A Book Apart

Erika Hall Reflects on 5 Years of Conversational Design

Apr 03, 2023

ABA logo.

Erika Hall eating a book-shaped cake with her hands.

Conversational Design celebrates five years in the world, Erika Hall shares her reflections on what’s changed since the book was first published, what we need to pay attention to regarding AI, and what to explore next.

A Book Apart: What is the most important thing practitioners must pay attention to as they put the advice in your book to work today?

Erika Hall: Obviously, generative AI, and specifically ChatGPT, have exploded into reality in the past couple of months. We’re going to be separating novelty and fantasy from utility for a while yet.

Keep this in mind, though, as you ride the wave:

“I think one thing about the current AI hype cycle is that 1) a lot of people are going to lose their jobs and 2) this technology will not actually be capable of doing those jobs, can both be true.”
Tobias Wilson-Bates, assistant professor of British literature

I’m starting to get a lot of questions about testing prototypes with AI instead of an actual representative human, which is more about my other book, but seriously, think about what you’re trying to accomplish with your work and what your standards should be. “Hallucinations” and misinformation aside, these aren’t sentient, self-reflective systems, and the responses are restricted to information gathered from its training dataset and conditioned by guardrails.

Some of the examples might seem quaint, but all of the principles stand.

ABA: How has the way you talk about the book changed since launch? Why has it changed?

EH: The purpose of the book is to keep humans at the center of the conversation, regardless of how technology changes. Maybe the division of labor between humans and machines will shift a little bit.

The core messages of listening before you speak, thinking of maintaining state in conversations that happen across modes and devices, and being cooperative, context aware, truthful, and ethical—are just as true if not more so.

The fundamental principles are the same whether you’re designing an interaction between two humans, between a human and a conventional website/app, or between a human and an “AI” agent. Organizations need to be consistently thoughtful across all of these scenarios.

Also, Twitter has even less of a coherent strategy than it did five years ago.

ABA: What kind of feedback have you gotten from readers? What’s been unexpected, thematic, or particularly poignant?

EH: I’ve heard that a lot of readers are positively surprised by the contents of the book because there is so much shallow thinking on the topic, and generally about technology-driven hype.

Just because an interaction looks or sounds like a conversation, doesn’t mean it functions like a conversation or is easier than clicking on some links.

So, organizations forget—over and over and over—that the whole point is creating more value, making a task faster and easier, etc. Solving a really hard engineering problem isn’t enough. You have to connect that problem-solving to human behavior in a real world context.

Just look at the trajectory of Amazon’s Alexa. Amazon poured billions into the technology based on the idea that masses of customers would shop by voice. I am positive there were some very smart researchers and designers in there whose insights were drowned out by one-track wishful thinking.



Want to explore even more about conversational design? Erika’s got you covered with her excellent recs to read, watch, and listen to:

  • Hard Fork is a newish technology podcast from journalists Kevin Roose and Casey Newton. They make sense of the hype in the headlines, not exclusively about generative AI but they’ve done a good job wrestling with the ramifications in real time. Listen ➜
  • Humans tend to switch modes pretty easily in conversation with each other. Designing across devices and modes is extremely challenging, especially given how organizations like to silo their design approach by device and mode. Cheryl Platz wrote a great book, and in this talk she summarizes some of its key points. Watch ➜
  • From New York magazine, this piece is written in an expansive narrative features style, but the profile of linguist Emily Bender offers plenty of nuggets worth following up on. Read ➜
  • I haven’t dug into the Conversation Design Institute too deeply yet, but I’m intrigued by some of their courses, like Introduction to Conversational AI Ethics and Conversational AI Workflow. Browse ➜