I came to America from Bangladesh at the age of 17 with little money in my pocket and no idea what kind of world I was walking into. What I had, and have relied on in every decision I make ever since, was the ability I developed early in life. It was the ability to understand what I actually thought and act on it, even if no one else agreed with me.
There is a word that describes that ability. judgement. And that’s what I’m most worried about losing right now.
The reason I’m interested is because I use AI every day. I use it to build applications, develop frameworks, design visual assets, and research what’s going on at the cutting edge of areas I need to understand. I use this to stress test discussions before bringing them to my team or clients. I use it to consider how a message will arrive before sending it. In terms of raw results, I’m more productive today than my team of 10 was five years ago. And it’s not just about speed. My work has objectively gotten better.
At this point, I’m still one of the early adopters. But in the near future, this will become a reality for every job. Just as few people today can choose whether to use computers, email, or the Internet, few people will have a choice whether to use AI or not.
So this is not an article about whether to use AI or not. It’s about how you maintain your judgment.
Core skills for successful use of AI
AI is very powerful, but it’s also powerful in the so-called general sense. Generative AI models can be trained on all the insights of all the sciences, all the works of the greatest artists and the brightest business thinkers, but they generally cannot and cannot know what is important. your specific situation. The machine doesn’t know what trade-offs you’re willing to accept, what your experience tells you about how something will actually land, or what the right decision is given all the things it doesn’t know. That knowledge is yours, and you can use it to turn AI’s general capabilities into useful ones. you.
Without user judgment, the AI provides fluent but common output. It allows you to get something that no one else has. And the combined power of AI and your judgment can be far more effective than using either alone. However, this is only if users actively participate in the conversation and think alongside the tool rather than following it.
This means that leveraging AI successfully isn’t fundamentally about creating better prompts or knowing which model to use. Rather, it’s important to stay actively engaged with what comes back. It’s that simple.
The principle may be simple. In practice it is not so simple. It becomes so easy for AI models to respond to requests and create outputs, resulting in a phenomenon known as cognitive offloading. Reducing our mental load is precisely the appeal and promise of AI. Delegating some tasks allows us to think more deeply and effectively about other things. But the risk is that we outsource the wrong things, that is, we outsource higher-level judgments about what’s important: the ultimate meaning of a piece of work or creative design. We will then stop using AI to support our thinking and instead rely on machines to do it for us.
How to effectively utilize AI
These are not rules to avoid using AI. These are exercises to get the most out of by making sure you’re always in the driver’s seat.
- Brainstorm your ideas before creating your prompt. You can’t effectively delegate to an AI model if you don’t have full control over the task. Start with a topic you’re interested in and ask the machine what it thinks about it, and you’ll start by following the machine. Instead, come with a development you want to protect. That doesn’t mean we shouldn’t have AI challenge our views or help us strengthen them. However, before you type the first word, you need to understand your strategic intent. That’s what makes you the designer of the output, rather than the agent of the AI.
- Define the destination before delegating execution. Don’t give your AI model a blank canvas. Give them a defined problem with a sketch of the destination and use it to fill in the details. That way, the architecture (requirements, dimensions, logic) remains yours. If you can’t explain what you’re looking for before asking a question, you’re not ready to use the tool. The result is an output that looks finished, but isn’t actually yours.
- Find research using AI. Interpret it for yourself. AI is very good at surfacing research, mapping fields, and presenting evidence that confirms or challenges views. Use it for all of that. But don’t accept the summary of the results it digs average. Your interpretation, where the evidence leads, what supports it, and where it falls short, must be your own.
- Be careful when reaching for AI to avoid unpleasantness. Sometimes we prompt AI because we want it to take some of the processing load off a task. Other times, we reach for unpleasant things just because we can easily avoid them. Discipline is identifying the type of difficulty that creates friction, leaning into it, and tracking where it leads. This is true whether you’re developing a business strategy, working on an article, or building a product.
human edge
The practices described above are not rules that limit how or whether AI may be used. These are the tools I use to allow me to stay in the driver’s seat rather than being a passenger being driven around by very powerful machines.
So a core skill in the AI era is the willingness to drive yourself. To be a designer or orchestrator of a process, rather than a ship tossed about by storms in a sea of mechanical reasoning. It means knowing where you are going before you set out and never letting go of your vision.
The most powerful technology ever built cannot save someone who has stopped thinking for themselves. The discipline needed in the AI era has two parts. It’s about learning how to use the system effectively while refusing to let it take advantage of you.
