Frequently Asked Questions
These common questions and their short answers are taken from Christopher Noessel’s book Designing Assistant Technology: AI That Makes Us Smarter (2026). You can find longer answers to each in your copy of the book, either printed or digital version.
- Is this about chatbots?
Lots of ink and phosphorus has been spilled writing about those. It’s a rich field and the coverage is warranted. I even have some favorites to recommend. Erika Hall’s Conversational Design, Cathy Pearl’s Designing Voice User Design, Diana Deibel and Rebecca Evanhoe’s Conversation with Things, and Robert Moore and Raphael Arar’s Conversational UX Design come to mind. So, I don’t feel the need to add to the excellent thinking there. I do speak to chatbots throughout the book, most directly in Chapter 4, “The Five Universal Assists,” under “Know,” but I intentionally do not go into any depth about them. - Why don’t you offer a definition of artificial intelligence?
I don’t think we need it. We have words like “flower” that have different casual and botanical definitions, but nothing breaks if the layperson speaks of a daisy as a flower and a fig as a fruit. And sure, what “counts” as AI has changed throughout the decades since the term was coined but so have most words. Even if “AI” has a loose and blurry boundary, we operate with vague categories all the time. For example, the “produce” section at a grocery store has more than produce, and I’ve never heard of a shopper freaking out about this fact.
So, let’s proceed with our broad, fuzzy, changing understanding of the term. I don’t think anything will break.
I do hope to encourage the reader to get crisp about seeing assistants and agents differently, so Chapter 3, “What Is ‘Assistant’ Tech,” should help tease that out. - What’s the most accessible example of an assistant technology you can give me?
Navigation apps are ubiquitous for smart phone users and one in particular was quite formative to my thinking on this topic. Chapter 1, “Google Maps as a Bejeweled Crutch,” goes into some detail about that app, and how it is well designed to help with the common task of navigation and the negative effect it can have. Chapter 2, “Will Assistants Doom Us to ‘Stupidity’?,” describes a simple intervention that not only mitigated that risk, but flipped it on its head. Navigation apps are my go-to example when discussing assistants. Chapters 5–9 contain dozens more examples. - I have an assistant product. What do I need to get started?
If your product is really all about assisting users, of course you’ll want to understand the Five Universal Assists described in Part II, “The Ways You Can Assist,” but importantly, I want to point you toward the chapters that describe the larger risks that providing assistants entail. I introduce that problem in Chapter 1, “Google Maps as a Bejeweled Crutch,” and discuss the ways to combat it in Chapter 2, “Will Assistants Doom Us to ‘Stupidity’?,” and again throughout Part III, “Mitigating the Risks.” - Why are people talking about agentic rather than agentive now?
Because they’re different things. Agentic refers to backend architecture, and that has gotten lots of buzz of late. Someone recognized the need for a categorical term and constructed “agentic.” I’m quite glad that this person didn’t know much about morphological etymology, or they might have reinvented “agentive” and relegated my work to the dustbins of search-engine optimization. As it is, these are two different, if related, things, and the recent popularity of agentic backends is getting people to rediscover agentive front ends, so despite the confusion, I’m OK with it. You can read more about agentic concerns for assistants in the sidebar in Chapter 7, “Plan.” - Why don’t you go into depth about interfaces?
Assistant technology differs primarily in use cases rather than interfaces, so Part II, “The Ways You Can Assist,” is dedicated to identifying and describing those, just as I did in the prior book, Designing Agentive Technology: AI That Works for People (Rosenfeld Media, 2017). Additionally, Chapter 12, “Cognitive Forcing Functions,” describes an additional interaction called “Human Goes First,” as well as other mitigation patterns that would need interfaces tailored to your particular domain. Readers can draw on the existing tools of interaction and interface design for best practices around individual touchpoints. - I saw an em dash in the text! How much of this book was written by AI?
There is a mind worm going around which says that human authors don’t use em dashes, and despite this, they somehow wound up in the models and are over-represented in text output by large language models (LLMs). But let’s be clear. Em dashes have been around a long time, long before LLMs. The quote that begins the last chapter contains two of them, and it was written in 1909. The Oxford English Dictionary website says the first em dash we know about came from the 1830s. There’s almost 200 years of human-made, typographic history there, and it doesn’t disappear just because you didn’t know about it. So, its mere appearance in a text does not mean it is AI slop.
I am well aware and highly supportive of the idea that modern readers should keep a sharp eye out for slop. And that an author of a book about AI might have a positive-enough disposition toward the technology that they might abuse it. So, I have added Appendix B, “AI Colophon,” at the end of the book that lists—by chapter, model, and some cases prompt—each use of a LLM that I performed as I went along. I conceived of this very early in the writing process, and I was diligent throughout, so you can see exactly how I used that technology. This book may be slop, but if so, it’s not because I asked an AI to write it. Chapter 14, “Assistants in an Era of General AI,” is a notable and explicit counter-example. It was pointedly cowritten with an LLM. - Do you have be so political? Isn’t this about the design of technology? Shouldn’t it be more neutral?
Technology and design is inherently political (AI even more so), and its veneer of neutrality is one of its greatest, most nefarious dangers. Especially now, with technology powering much of the global disinformation machine, and everyone reeling from the dual impact of fascism and AI, it is imperative that we confront the politics of these
things in every platform available to us. To do less is to side with oppressors, and if you haven’t noticed, there is a lot of oppression going down. - You’re just another cheerleader for the future, blithely bringing artificial intelligence doom down on us all! Wake up, sheeple!
Sharp-eyed, long-time readers will notice that this same question appeared in the FAQ of Designing Agentive Technology. But here, I have a slightly different answer. Because I am more worried now than I was then.
As predicted for *checks notes* decades now, the power and reach of AI is accelerating in hockey-stick fashion, and “it’s-just-business” types are replacing workers with jobs and asking everyone who’s left to do much more with much less. If the notion of agents has emboldened those its-just-business decisions, I apologize. I certainly wrote my book with the intent to empower people, not to equip others to replace them. I trust others are similarly worried about the upshot of replacing all those people with AI. Is this the society we want? What do we do to curtail its worst effects?
While most of those answers may be political, consider this book as a kind of design antidote to the misuse of those concepts. Because assistants are meaningless without someone to assist, I’m in favor of technology that empowers people, rather than empowering wealth aggregation machines that exacerbate inequality and the return to feudalism. We’re at yet another threshold where technology stands to restructure our societies, and we should steer that restructuring in as humane a way as possible. I hope this will help.
