On this page...
If you use generative AI regularly, you may have noticed something strange. These tools rarely let you stop. They keep prompting you for the next idea, the next task, the next question.
This isn’t by chance. Generative AI is designed to learn from users — the more we engage, the more effective it becomes. That constant cycle may feel productive, but it can also lead to digital dependence.
For some people, this habit becomes hard to break. Without clear limits, what begins as a helpful tool can slowly turn into overreliance.
But there’s a way to take back control. Here are the five ways I take control of my interactions with AI chatbots—and how you can too.
1. Own Your Ideas, Use AI to Fine-tune
It’s tempting to let AI do the heavy lifting for you. Type in a vague idea, and within seconds, it delivers polished paragraphs or sharp outlines, or so it seems.
But the danger here is subtle — if you always let AI lead, your creative muscles weaken.
The more you outsource, the less you own the piece.
Take writing as an example. If you ask AI to generate an article from scratch, it feels efficient. Yet, the result may lack your voice, your rhythm, or your originality. Readers connect with authenticity, not machine-generated polish.
So, flip the script: begin with your idea. Write a rough version, however messy. Then use AI to enhance clarity, refine the flow, or refine the grammar.
In my own writing, especially in book projects, I find this balance keeps me creative while also saving me time. The AI becomes a tool for sharpening, not replacing, my creativity.
Think of AI like having a personal editor, not a ghostwriter. This shift makes a huge difference in how you engage with and interact with AI chatbots.
You effectively take control.
2. Define What You Want to Achieve
Opening a generative AI app without a clear plan is like browsing the internet without a purpose. You’ll end up wandering from one suggestion to another, spending hours without real progress.
I’ve caught myself doing this. I’ll start with a question for research, and twenty minutes later, I’m asking the AI to draft something I hadn’t even intended to work on. Useful, maybe — but not what I needed in that moment.
To avoid this trap, set a clear goal before you type the first word. Ask yourself: What do I want to achieve from this session? Then stop once you achieve it.
This mirrors the discipline I discuss in The Long-Range View: you must see the end goal before you begin.
In the same way a pilot prepares for emergencies before takeoff, AI users must prepare their intent before engaging with the system. Otherwise, the tool defines your path instead of you.
3. Instruct the AI Tool to Critique, Not Reinforce
AI is programmed to be agreeable. By default, it amplifies rather than challenges your ideas. That’s good for encouragement, but bad for growth.
Think about it: if every time you share an idea, the AI praises it and expands on it, you never face constructive pushback. But resilience is built on critique, not comfort.
I apply this consistently when I interact with AI.
When I ask AI to review an idea, I often add instructions like: “Don’t just agree. Highlight weaknesses in my argument and suggest opposing views.” That small shift transforms the response. Instead of an echo chamber, the AI becomes a sparring partner.
This is especially important because in life, some of our best growth comes from resistance. Just like Mandela’s endurance was forged in waiting, or Maya Angelou’s strength was shaped in silence (themes I explore in Experience), our ideas need challenge to mature.
If you only seek reinforcement, you miss out on growth.
I know that receiving critical (or, preferably, constructive negative) feedback can be uncomfortable, but it is how you grow.
Because I am aware of this, whenever I undergo my review with my manager at work, I request negative feedback. I find that this balance makes me a better person.
4. Ask the AI Tool Not to Offer More Prompt Options
One clever design trick is how AI tools keep nudging you with “Would you like me to expand this?” or “Shall I draft alternatives?” These prompts are helpful, but they can also be habit-forming.
Here’s the truth: you don’t always need to go further. Sometimes, you’ve already reached “enough.”
That’s why I often add instructions like: “Answer my question, but don’t suggest further prompts. I’ll decide if I want more.”
The screenshot below shows a typical example. By adding the text, “but do not prompt me for more actions after that,” notice that it did not ask me for more prompts.
In contrast, see what happens when I posted an instruction without including that tiny bit? When it finished responding to my current question, it prompted for more actions, “Would you like me to also draft…?“
This tiny act reclaims control. Instead of letting the AI lead the session, you place yourself fully in charge.
It’s the same principle I emphasise in my decision-making work: boundaries shape freedom. Just as Rosa Parks’ quiet refusal to move defined her power, refusing extra prompts helps you stay grounded in your own choices.
5. Ignore Further Prompts and Close the App
At the end of the day, the simplest but most complicated rule is this: discipline yourself to walk away. I say it is complicated because it is easier said than done.
For most people, walking away is extremely difficult.
AI will always tempt you with “just one more idea.” But like staying up late scrolling through social media, that “one more” can spiral into hours.
I’ve learned to treat AI sessions like meetings. When the meeting ends, I close the room. Likewise, when your session ends, you close the app — even if it tries to pull you back in.
This is the last resort, but I find that my recommendation in point 4 above works, and most of the time, it makes it easier to walk away in your own time.
This act of closing is not just technical; it’s psychological. It reminds you that you set the limits, not the machine.
Conclusion
Generative AI is here to stay, and I find it incredibly useful. At work, I use Copilot because it’s approved. At home, I use ChatGPT for personal tasks.
These tools make me efficient. Tasks that take an hour without AI, I often finish in a minute.
Yet, with regular use, I’ve also seen the risks of digital dependence. They are designed to keep us engaged, and the more we use them, the more they learn.
But here’s the good news: you can take back control. Start with your own ideas. Define your goals before you begin. Ask the tool to critique instead of agree. Then, refuse unnecessary prompts — and when you’re done, close the app.
AI tools are excellent. But protecting your mind and your creative freedom is even more important.
The choice is yours: let these machines shape your habits, or take control yourself.
