A Note on AI Assistance
I have made extensive use of artificial intelligence tools in writing and revising this book and developing the accompanying software, visualizations, and simulations. That assistance includes suggesting and revising prose, implementing and debugging code, preparing figures and worksheets, checking consistency, and reviewing the alignment of exercises with learning objectives. It has shaped the work you are reading, so it deserves a direct acknowledgment. The responsibility for what I publish remains mine, including its errors.
How I Work with These Tools
My practice is generally to treat an AI assistant as a capable, fast-working junior colleague who needs direction and whose work needs review. I begin with notes about what I am trying to accomplish, the steps required, and the difficulties I expect to encounter. For prose, I usually develop a first draft before asking the assistant to read it alongside those notes. I ask it to identify problems in reasoning, pacing, structure, terminology, and expression, then discuss the findings before authorizing revisions. For software and learner resources, I also use it to implement changes under my direction and help build checks of the resulting behavior.
I keep a separate guide to the voice I am trying to achieve and ask the assistant to use it when revising prose. A technically acceptable sentence can still flatten a joke, blur a distinction, or replace an argument with something that merely sounds respectable. Those are losses, even when the new sentence is smoother.
I also invite the assistant to ask clarifying questions. This has been one of the most useful parts of the process: sometimes a question reveals a distinction or dependency I had not considered. After revisions, I try to put the work aside and return with fresh eyes. There are usually more changes to make: sharpening an explanation, adjusting a metaphor, or putting a joke back where it belongs. AI assistance has shortened the distance between a reasonable first draft and a finished piece of work; it has not removed the need to travel that distance.
What I Have Found Useful—and What Needs Scrutiny
In my experience, these tools are particularly useful for highly structured tasks: repetitive formatting, boilerplate code, a visualization with clearly specified data, or a consistency check across several documents. They can also be useful partners in diagnosing bugs, suggesting revisions, and identifying gaps in an explanation. The result depends heavily on the instructions, the context, and the particular task. These are observations about my working practice, not a permanent ranking of what every AI system can do.
The difficulties become more serious when the task requires sustained judgment across a chapter, a curriculum, or a software system. An assistant can produce a plausible explanation that misses the point, a polished paragraph that no longer says what I meant, or a calculation that quietly uses the wrong boundary. A proposed source or quotation still needs checking against the source itself. Code that runs still needs to perform the intended operation. A favorable AI review is not independent evidence that a learner can use the material successfully.
The same distinction matters for the examples in this book. AI assistance in preparing a case, calculation, or simulation does not turn it into an observed business result. The fictional Wash n’ Fold case remains instructional fiction; its records and modeled changes must be read within the assumptions stated in the chapters.
Keeping Hold of the Work
AI is a little like improv. The improviser’s motto is “Yes, and . . . ” which is brilliant for keeping a scene moving because the performer accepts what has been offered and elaborates upon it. In the same way, an AI assistant can be very good at keeping movement going, including when the premise should have been questioned several paragraphs ago. That can be a fun way to explore an idea. It can also produce a horrible tangled mess of nonsense if nobody stops to ask what the work is supposed to establish.
I think these tools can be fantastically useful when you know enough about the subject to direct the work and examine the result critically. If you ask them to substitute for knowledge you do not have, their fluency can make the problem harder to notice. The more the pieces need to hang together—as a section, a chapter, or a working system—the more carefully the author has to specify and inspect their connections. That is the discipline I have tried to follow here.