You use AI to write and so does your boss, your student, and everyone else. That ship went arguing that the ride was a waste of air.
The real question is: how do you use AI?
AI is a fork in the road disguised as a shortcut. Under one method, a coach learns your weaknesses, challenges your thinking, and pushes you to your limits. On the other hand, it’s a ghost, and you put your name on someone else’s thoughts, slowly forgetting that you ever had your own.
The tension between these two approaches is a clear challenge of our time. Psychologists call our mental overload a ghostwriter, and we’ve been doing mild versions of it forever, from writing a shopping list to keeping contacts on the phone. But with AI we no longer load gossip. We release the thought itself. And easy to do so is a temptation that people struggle to resist.
The easiest way: Downward rotation
When you ask ChatGPT to write your email or memo, you’re not just putting in the time. Skip the mental exercise. And like any muscle left idle on the couch, the brain responds accordingly: it atrophies.
This simple method creates a vicious cycle.
First, AI addiction: the more we use large-scale linguistic models (LLMs), the harder it is to stop. Researchers in the Annals of the New York Academy of Sciences describe dependency as two things: functional dependency, where we rely on AI for productivity, and existential dependency, a deep psychological attachment that involves self-awareness, emotional regulation, and even friendship. Unlike our relationships with calculators or spell checkers, LLMs involve higher levels of cognition: analysis, synthesis, persuasion, judgment.
Second, we become weaker, and the gap widens. A Microsoft study found a significant negative correlation (r = -0.49) between the frequency of AI tool use and critical thinking scores. The MIT Media Lab went one step further, strapping EEG monitors to participants’ heads: ChatGPT users showed weaker brain connectivity than any group tested. They struggle to remember the content of the stories they just “wrote”. The tool did the thinking; a person was simply holding the steering wheel of a self-driving car. Meanwhile, AI models are improving every quarter. The distance between what we can do without help and what a machine can do widens like a crack in a windshield, spreading silently until everything falls apart.
Third, mental commitment: the act of accepting the results of AI is almost zero. In a series of experiments involving more than 1,300 participants, Shaw and Nave found that regular AI users stopped evaluating the AI’s work. When the AI was wrong, they were. They had given not only the task of thinking but even the responsibility of verifying it. The better AI gets, the more we agree.
This vicious cycle inspired me to write the following Zen Koan about AI:
Learn without learning.
Write without writing.
Think without thinking.
Can you be wise by being dumb?
I do not think so.
Another way: AI as a trainer
So much for a ghost writer. Let’s talk about the coach.
The same tool that threatens to dull your cognitive abilities, if used with discipline, can sharpen them instead. The key change is in your intent and mindset: use AI to improve the quality of your work, not just quantity or speed.
Here’s what it looks like in practice.
Think, don’t delegate.Use AI as a sparring partner for ideas. Push it to generate multiple angles on the topic, then counter its suggestions. The goal is not to accept what you give out but to fight it, to let the conflict create ideas between you that you would not have produced on your own.
Pressure-check your arguments.Ask the AI to find holes in your thinking, strengthen the opposition, to point out the argument you’ve been avoiding. This is the mental equivalent of hiring a boxing coach that fits you.
Build a skeleton yourself.Create your own frame and message before consulting the machine. Argument structure is where most of the real thinking resides. If you take out the plan, you are left decorating someone else’s house.
Strengthen your research.Use AI to reveal related work, nearby fields, and sources you might have missed. Let it expand the radius of your awareness without changing judgment about what is important.
Polish.Let AI improve your grammar, reinforce your pronunciation, and flag obscure passages. This is the digital equivalent of a copy editor, a role that enhances the writer without replacing it.
Improve clarity.Ask the AI if your prose is doing what you intended. Is the conflict going down? Does the structure make sense? This turns the machine into a mirror that reflects your thinking back to you with helpful annotations.
Review everything.This is non-negotiable: Read your output with the editor’s skepticism, not the customer’s gratitude. Check the facts. Validate claims. Make sure the word is yours. If you can’t explain and defend every sentence, you haven’t written an article; registered one.
Finally, learn from the process. After each project, assess what the AI revealed about your weaknesses and strengths. Has it been improving your conversions? That tells you something. Does it catch any logical gaps that you missed? That tells you something, too. Treat each interaction as a lesson in your cognitive blind spots, and to celebrate your strengths.
This article was written using the exact process I described. I used AI to think, pressure-test my argument, hunt down research and statistics, and improve my prose. But the thesis is mine. The building is mine. Words, metaphors, beliefs, and errors are my own. I reviewed the entire claim. I rewrote the broken AI verses. I cut through proposals that were technically smooth but intellectually empty.
It took longer than giving the topic to the LLM and telling it to “write,” but that’s exactly the point. Every time you open a chat window, you’re stuck at the fork again. AI trainer or AI ghostwriter.
Editor’s Note: GeekWire publishes guest comments to encourage informed discussion and highlight diversity of opinion on issues that shape technology and the startup community. If you’re interested in submitting a guest column, email us at [email protected]. Submissions are reviewed by our editorial team for relevance and editorial standards.
