I Don’t Need AI for That

Sometimes I give a coding task to an AI agent and watch it write code I know I could have written myself. It might take me longer, but I could do it. I might even do it better because I understand the project, the users, and the reasons behind the decisions.

Instead, the agent writes the code while I open TikTok.

Technically, I have saved time. But I’m not sure what I saved it for.

This is the uncomfortable part of AI for me.

I work in technology and use these tools regularly. They help me code, brainstorm, search documentation, and work with hardware that I’m still learning.

I’m not interested in pretending they are useless or that I’m going to remove them completely from my life.

But I’m becoming more careful about what I give them.

We were capable before this#

Only a few years ago, we wrote our own emails and Slack messages. We planned our own days, read documents, researched problems, and worked through ideas without asking an assistant to do every step.

These things were not considered special skills. They were simply part of being a person.

Now AI is being added everywhere. It can summarize every document, compose every message, plan every day, generate every post, and suggest what we should say next.

The argument is usually that this saves time. Sometimes it does.

But efficiency is not the only thing that matters. Writing a message is also an act of thinking. Choosing how to explain something requires you to understand it. Reading a document forces you to decide what matters. Planning your day means considering what you actually value.

When we outsource all of that, we are not only removing work. We may also be removing participation.

Becoming a spectator#

The danger I feel most personally is numbness. When an AI agent produces something for me, I can still review it and approve it. But reviewing is not the same as creating. If I do this repeatedly, I begin to feel like a spectator to my own work.

The same applies to creative work. AI can produce posts, images, videos, and endless variations of an idea. Some of them may even perform well. But if I did not struggle with the idea, make the decisions, or put some part of myself into it, the result rarely feels like mine. It may be content, but it does not feel like creation.

Something similar happens with expertise. Expertise is not simply having the correct answer available. It is knowing how to reach it, noticing when something is wrong, understanding the context, and being able to make a judgment when the instructions are incomplete.

If I blindly accept whatever an AI produces, I may get an answer without developing any of those abilities.

Eventually, I might know how to ask for things but not how to do them. I might be surrounded by information without being able to process it properly.

AI can still help me think#

This does not happen every time I use AI. I’m relatively new to hardware, for example. AI has helped me understand wiring, navigate technical documentation, and ask questions that would otherwise take much longer to answer. In that situation, it can help me learn alongside the work.

It is also useful for searching large collections of documentation. I can ask a question, find the relevant section, and then process the information myself. It helps me reach the material without necessarily replacing my understanding of it.

Brainstorming can work in the same way. AI can expose me to another direction or help me see something I missed. But I still need to decide whether the idea is useful, whether it makes sense, and whether I believe in it.

The distinction is not simply between using AI and avoiding AI. It is between participating and outsourcing. Am I using it to extend my thinking, or to escape thinking altogether?

Think first#

The rule I want to follow is simple:

Rely on my own intelligence first.

Before asking AI, I want to research. I want to look at what real people are doing and what they respond to. I want to ask whether I would care about the thing I’m creating.

If I’m building something, I want to think from the user’s perspective. Not only, “What did they ask for?” but, “Why do they want it? What are they actually trying to solve?” That human work needs to happen before the prompt.

AI can help me explore a solution, find information, or work in an area where I’m still learning. But it cannot replace caring about the person on the other side. It cannot decide what deserves to exist. And it should not automatically provide my opinions, personality, or voice. Perhaps the practical test is this:

If the thinking, learning, or creating is the meaningful part, I should not outsource it.

I can write my own emails. I can speak to people myself. I can plan my own day. I can create something that is imperfect but actually belongs to me. My own intelligence is still enough for far more than the current excitement around AI would have me believe.

I’m not aiming for an AI-free life. I want a life in which AI remains a tool rather than becoming my replacement. Less AI where it makes me a spectator. More AI where it helps me learn. And, overall, more human.