What Are People Deciding, Not Just Doing?

I've been thinking a lot about AI lately and where it fits into the work we do. Like many people, I've been experimenting with it. Personally, I use AI for collaboration, to help me work through a thought or an idea. I watched a MasterClass by Allie Miller on AI, and she summed it up by saying AI takes you from 0 to 30. You take it to 60.

Out of curiosity, I asked AI "How do you decide? Or, do you?" It responded with:

"That's a fascinating question.

My honest answer is: I don't decide the way you do.

I can evaluate options. I can compare tradeoffs. I can identify patterns. I can predict likely outcomes based on information I've seen before. From the outside, that can look a lot like judgment.

But there's an important difference.

When you're thinking through the balloon column example, you're doing something I cannot do in the same way. You're pulling from years of lived experience, including things that were never formally recorded.

You might think: "The customer said outdoors, but the way they hesitated makes me think they haven't fully considered the weather." Or -

"This venue reminds me of a job from ten years ago where we ran into a visibility problem." Or - "Technically Method A would work, but I've got a nagging feeling Method B will create a better customer experience."

What I do is closer to this: Given the information available, which patterns seem most applicable?"

Fascinating.

I sat with that for a moment. The more I thought about it, the more I realized that many of the decisions I make every day don't feel like decisions at all. They're habits, preferences, adjustments, and lessons learned from experience. Most of the time, I don't even notice I'm making them. That's when an analogy of baking a cake came to mind. On paper, it was simple: cake flour, egg, baking powder, sugar, salt, oil, and milk. Follow the directions and you get cake. But I can't eat egg protein, so right away I'm making a substitution. Instead of an egg, I use ground flaxseed mixed with water and let it sit until it becomes gelatinous. That part is easy enough. Plenty of recipes and AI tools can tell me how much flaxseed and water to use. I also tend to replace oil with unsweetened applesauce, and because applesauce adds sweetness, I know from experience that I can usually reduce the sugar a bit. How much? Well, it depends.

Where the recipe tells me what to do, experience helps me decide what to do when the recipe needs tweaking to fit my situation. The more I thought about it, the more I wondered if businesses run into the same challenge.

That's where things get interesting. If I ask AI, it can give me guidelines and explain the science. It can suggest substitutions. Sometimes it gives me ideas I wouldn't have considered myself. But very quickly we leave the world of process and enter the world of judgment. What if I don't have applesauce? Can I use pear sauce? Should I reduce sugar?  What if I'm baking this for company and care more about texture than calories? What if it's winter and the air is dry, or August and the humidity is through the roof?

At some point, it's no longer about the correct answer. There may be several choices that can work well in my cake recipe; none of them are wrong. The challenge is deciding which option is best for this particular cake, with these ingredients, in this kitchen, on this day. The recipe tells me what to do. Experience helps me decide what to do when the recipe no longer fits the situation. The more I thought about it, the more I realized that "best" is often a judgment call.

Most discussions focus on documenting the process. Map the workflow. Define the steps. Create the procedure. Build the automation. All of that has value. But I wonder if we’re sometimes documenting only part of the work. I mean, a process tells us what people are doing. Judgment, on the other hand, determines which process to use in the moment.  

A person with experience is often evaluating factors that may not be written down in a SOP. They're noticing exceptions and making adjustments. They are responding to situations no one anticipated. Many of those decisions happen so naturally that people don't even realize they're making them.

Ask someone how they do a task and they'll describe the steps. Ask why they chose that approach instead of three other reasonable alternatives, and now you're getting closer to the expertise. Now, you're not just documenting the process, you're also uncovering the judgment.

When everything is routine and predictable, process works beautifully. That's where AI can shine. It can process information faster than any human ever could. But when ambiguity enters the picture, things become more interesting. Humans bring judgment to ambiguity. The requirements are changing. The information is incomplete. The exceptions start piling up. Someone has to weigh the tradeoffs and determine what is most appropriate given the circumstances.

I suspect a lot of organizations exploring AI would benefit from asking a simple question before automating a workflow: What are people deciding, not just doing? The answers may reveal opportunities for automation, simplification, or process redesign. But they also reveal where experience, context, creativity, and human judgment are still doing heavy lifting no one has bothered to document. And perhaps that’s the real starting point for AI preparedness. Not identifying the steps first. Identifying the decisions.