Why Your AI Results Aren't Working (And It's Not the Tool)

I get asked some version of this question almost every week: “Which AI tool is the best one to use?” While some people may have different motives for asking, I feel most people ask it out of frustration.

They are using AI, and their results feel inconsistent, generic, or just not useful for their situation. The natural response, then, is to assume they are using the wrong tool. In most cases, that’s not what is happening.

The truth is a little harder to hear: “You’re using the tool in a way that all but guarantees weak results.” Which is shocking to most people, because most people are still using AI at its most basic level—what I call the “Google level.”

They think of something they need, they type a question, they read the response, and they decide whether it’s good or bad. That’s it: one question, one answer, done. That’s where it breaks down.

The problem is, AI isn’t Google and doesn’t work well in that structure.

It will give you something, and sometimes it will even look polished. However, if you’ve ever read a response and thought, “This is close, but I can’t actually use this,” you’ve already experienced the result of an incorrect approach.

There are a few things happening behind the scenes that most people never account for. In our curriculum, we refer to them as the “Principles of AI,” and we divide them into two groups. The first are the principles that govern how AI operates, and the second are the principles that govern how we interact with AI. Knowing and applying a few of them can help move you away from surface-level AI interaction.

One of the biggest eye-opening principles for people is this: AI fills any undefined space in the work you give it. This is a big one because it essentially tells us that AI will fill any of the “gaps” we leave open.

If your request is vague, it doesn’t ask you to clarify. Instead, it infers missing details, and those assumptions are often where things start to go off track. General questions with no details are going to have inferences and assumptions built into the response due to the nature of AI. If you’ve ever wondered why AI gave you something you didn’t ask for, or included something in the output you didn’t request, you were not following this principle.

Another big principle of AI is this: If you don’t define how AI participates, it will. To clarify, this principle is not saying AI has “agency”; it is simply reinforcing the one we just discussed. If you don’t specifically tell AI what its role is, it will assume one, and it will do so using generic patterns or learned behaviors from past interactions and underlying algorithms.

If you are just dropping statements, inquiries, and general problems into the tool with no specified role, then you are leaving it up to patterns and probabilities to determine your response. The tool is operating without clear direction. It’s like playing a game of horseshoes and just trying to get “close enough.”

Interaction is iterative: This is the third principle we need to cover. AI is built differently than Google. It is designed to operate through interaction—specifically the back-and-forth that is a hallmark of iterative thinking.

The first response you get from AI is almost never the best response. Why? Because the tool is built to operate best in an iterative environment. This is why AI is often described as a “thinking tool.” The iteration between you and the AI is what refines the output and makes it usable.

Until you start reviewing AI outputs and refining them—adding additional context, clarifying what it didn’t take into account, specifying its role more clearly, and being explicit with the task and execution—you will stay at surface-level AI usage.

Finally, the last principle we will discuss is this: Human judgment remains responsible for action. Your judgment still matters. A lot.

AI’s effectiveness depends on human involvement throughout the process. It can generate options, structure ideas, and accelerate your thinking, but it cannot replace you. It does not replace your experience, your judgment, your values, your sense of right and wrong, or your instinct that only comes from actually doing the work. If something feels off, then do something about it. The output won’t fix itself.

AI is a tool... it may very well be a tool that changes the course of humanity. Either way, if you’ve been using it and your results feel inconsistent, frustrating, or underwhelming, it’s worth asking yourself a different question: not “Which tool should I use?” but “Am I actually working with this tool correctly, or am I just asking questions?”

If you start using AI correctly—like a process instead of a shortcut—the results tend to change pretty quickly.