This is a good example of someone using AI as a bicycle to improve efficiency in their own efforts rather than using it as a train where you tell the AI to do it all. Further expansion on that thought here - https://ombreolivier.substack.com/p/ai-bicycle-or-train?r=7yrqz
Just for fun I ran the two immigration paragraphs - yours and ChatGPTs - through pangram and it told me that yours was 100% non A! written while ChatGPT's was 100% AI written, which is further confirmation that pangram is pretty good at AI detection
"And the writing process does more than just produce a piece of writing. I learn a lot from researching, outlining, drafting, and revising, and the knowledge and worldview that is built up in my head through that process is part of the outcome. What I learn helps me decide what to research and write about next, serves as examples in future essays, and provides the substance of interviews that I give. Even if I could crank out perfect essays using AI, I couldn’t be a public intellectual that way."
This resonated with me because I've found that writing is often how I figure out what I think.
I use AI extensively for research, fact-checking, finding sources, testing arguments, and surfacing counterarguments. In fact, for longer essays I've written on India, institutions, and economic development, AI has become an extraordinarily useful research assistant. But when I'm wrestling with an idea, the act of turning a vague intuition into a coherent argument is usually where the real intellectual work happens.
Your post reminded me of Tomas Pueyo's recent essay arguing that most intelligence is really a search problem. I think he's broadly right. Very few of us are generating completely novel ideas. We are searching a vast body of facts, examples, stories, and arguments, then assembling them into something useful.
But if intelligence is search, the highest-value part may not be retrieving facts. It may be deciding what to search for in the first place.
The strength of your immigration paragraph wasn't the prose. It was the judgment. You noticed certain facts, ignored others, and linked them together in a way that supported a broader argument about progress. Another writer, presented with exactly the same evidence, might have chosen completely different examples and reached a different conclusion.
That's why I increasingly think the scarce resource isn't writing ability, or even information. It's the ability to notice. To spot patterns. To identify which facts are signal rather than noise. To make unexpected connections between ideas that initially appear unrelated.
As someone who writes primarily to clarify my own thinking, I've noticed that the finished essay is only part of the output. The other part is the mental model that gets built during the process. That model then influences what I notice in the future, which questions I ask, which facts seem important, and which analogies come to mind.
Perhaps that's why many writers remain reluctant to hand over drafting. Not because the machine can't produce acceptable prose, but because drafting is one of the ways they train the faculty that ultimately matters most: deciding what is worth noticing, and how seemingly disconnected observations fit together into a worldview they are willing to sign their name to.
Love this post! You've put in words my exact mindset as well. Use LLMs for researching, but the writing has to be my own words. I've been journaling since I was 11 years old and I've had a personal blog since 2008. Never needed any "tools" for writing besides my own mind and imagination, and I believe that's how it should be. I never wrote to "create content" since the old Internet was never about that, but still generated some deep & meaningful connections with strangers online. I'm getting nostalgic and I know the old Internet era is pretty much dead, but I still believe that the humanity can be kept alive by preserving our core methods of self-expression (writing, art, poetry etc.).
I think it’s funny that you use numbers and logical thought in your paragraph, while the AI uses highly emotional words and appeal, i.e. “absurd” in its. Kind of the opposite of what we were always told to expect,
This is a good example of someone using AI as a bicycle to improve efficiency in their own efforts rather than using it as a train where you tell the AI to do it all. Further expansion on that thought here - https://ombreolivier.substack.com/p/ai-bicycle-or-train?r=7yrqz
Just for fun I ran the two immigration paragraphs - yours and ChatGPTs - through pangram and it told me that yours was 100% non A! written while ChatGPT's was 100% AI written, which is further confirmation that pangram is pretty good at AI detection
Important point:
"And the writing process does more than just produce a piece of writing. I learn a lot from researching, outlining, drafting, and revising, and the knowledge and worldview that is built up in my head through that process is part of the outcome. What I learn helps me decide what to research and write about next, serves as examples in future essays, and provides the substance of interviews that I give. Even if I could crank out perfect essays using AI, I couldn’t be a public intellectual that way."
This resonated with me because I've found that writing is often how I figure out what I think.
I use AI extensively for research, fact-checking, finding sources, testing arguments, and surfacing counterarguments. In fact, for longer essays I've written on India, institutions, and economic development, AI has become an extraordinarily useful research assistant. But when I'm wrestling with an idea, the act of turning a vague intuition into a coherent argument is usually where the real intellectual work happens.
Your post reminded me of Tomas Pueyo's recent essay arguing that most intelligence is really a search problem. I think he's broadly right. Very few of us are generating completely novel ideas. We are searching a vast body of facts, examples, stories, and arguments, then assembling them into something useful.
But if intelligence is search, the highest-value part may not be retrieving facts. It may be deciding what to search for in the first place.
The strength of your immigration paragraph wasn't the prose. It was the judgment. You noticed certain facts, ignored others, and linked them together in a way that supported a broader argument about progress. Another writer, presented with exactly the same evidence, might have chosen completely different examples and reached a different conclusion.
That's why I increasingly think the scarce resource isn't writing ability, or even information. It's the ability to notice. To spot patterns. To identify which facts are signal rather than noise. To make unexpected connections between ideas that initially appear unrelated.
As someone who writes primarily to clarify my own thinking, I've noticed that the finished essay is only part of the output. The other part is the mental model that gets built during the process. That model then influences what I notice in the future, which questions I ask, which facts seem important, and which analogies come to mind.
Perhaps that's why many writers remain reluctant to hand over drafting. Not because the machine can't produce acceptable prose, but because drafting is one of the ways they train the faculty that ultimately matters most: deciding what is worth noticing, and how seemingly disconnected observations fit together into a worldview they are willing to sign their name to.
Love this post! You've put in words my exact mindset as well. Use LLMs for researching, but the writing has to be my own words. I've been journaling since I was 11 years old and I've had a personal blog since 2008. Never needed any "tools" for writing besides my own mind and imagination, and I believe that's how it should be. I never wrote to "create content" since the old Internet was never about that, but still generated some deep & meaningful connections with strangers online. I'm getting nostalgic and I know the old Internet era is pretty much dead, but I still believe that the humanity can be kept alive by preserving our core methods of self-expression (writing, art, poetry etc.).
I think it’s funny that you use numbers and logical thought in your paragraph, while the AI uses highly emotional words and appeal, i.e. “absurd” in its. Kind of the opposite of what we were always told to expect,
I hope the BBI addresses this to some degree! I could use the help, the guidance and I'm sure I'm not the only one.
There was some research recently that showed that college admissions essays are becoming more similar to each other post-GPT.