This piece highlights precisely the enormous gap I encounter all the time between intellectually mature and pre-internet educated adults (along with a very few unusual, and precocious, young people) and the vast majority of students in school, K-bacc.
The distinction between active AI use by people who already know how to conduct research, how to ask questions and pursue sustained inquiry, how to evaluate what the tools generate, and especially that asking questions and evaluating responses and results are even things humans do at all, on the one hand, and passive AI use by (currently younger) people who have never learned any of these things and for whom LLMs are simply ATMs with more buttons and steps, makes all the difference.
My older colleagues who love using AI tools and gush about all the new stuff they can do, if they are not also people who teach, especially people who teach students from backgrounds that don't bring with them a culture of reading, or researching, or reasoning, find themselves *completely unable to imagine* what it would be like to use AI e.g. for a writing project without being able to rely on what they already know, on the habits of inquiry and analysis and persistent evaluation developed over a lifetime of upbringing and schooling that forced this development, and offered only isolated shortcuts, like the calculator. They cannot get their heads around what it's like to treat every assignment like a new banking or transit app, just another thing to click on until you get the thing that will get you what you want, namely, in this case, the A. They can't see that without the experience, and the appetite, for the discovery and successful evaluation process, merely adding more steps to the assignment that *mimic* this discovery and evaluation just makes for more clicks that instruct the AI tool to itself *mimic* these processes and preserve them in the output a.k.a. the completed assignment, whether essay, or journal, or project, or poem.
The analogy to the advent of browser-based search engine internet experience actually bears this out, and distinguishes both that advent as well as the new AI tools from the emergence of writing as Plato experienced it, not least because of the staggeringly different timescales: we ourselves are the beneficiaries of a couple of millenia worth of adjustment (in the west) to the losses of oral memory culture and the benefits of writing, benefits that were and are still the subject of laborious evengelicalism in education to reach everyone in the classoom. With the browser- and search engine-internet, we're talking at most 30 years, not even, really, for general exposure and dependency. In my own experience over thirty years of teaching post-secondary students from all kinds of backgrounds I have found that we have--or had--barely begun to make the adjustment the OP here analogizes to the historical response to the advent of writing. That adjustment itself has, in my view, now been completely overrun by the arrival of LLMs and their universal adoption by the vast majority of current students.
When I hear about AI use *for students* that forces, de novo, rather than *relies on*, the development of habits of inquiry and evaluation, as well as actual, non-mimetic, knowledge, I'll be all ears. Until then, I cannot see that this tech is anything but harmful, in general use, for people coming of age intellectually from the turn of this millenium on.
Card Catalog's mission is an outstanding model and guide for thinking about these issues, and for bringing the fruits of our written millenia to an AI world--but it is crucial, in my view, to distinguish in this way between users of AI, because them has to have ears to hear.
Hana, your writing is advanced course work. Solid gold. Thank you so much for this substack. As others have mentioned, I will be referring to it often. And the readers' comments are just as valuable; like sitting in a class with graduate level thinkers. It's worth every penny of a paid subscription.
Thank you for posting this information. I attended school, university, and medical school when critical thinking was still embedded in classes. And have been hesitant to use AI or to recommend it to others, having read about "AI brain rot." Now that you have illustrated how it can be used properly, I'm eager to try it.
In it I used a metaphor I found the basis of somewhere else about AI as a bicycle vs AI as a train. The more I think about it, the more I like this metaphor and it seems consistent with both the scientific evidence you cite and your advice.
AI as bicycle means AI making you more efficient, the way a bike allows you to pedal and travel much faster/further/longer than you can running.
AI as train means you let the AI do the whole job of taking you to the destination for the price of a ticket.
When I upload a document, particularly a long document, I often use a prompt like "Please read this document in logical chunks. Create compact internal summaries for each chunk, then synthesize across them so no one section dominates. Focus on key themes, contradictions, caveats, and implications, and especially on [specific question I have about the document]."
This matters because an AI can focus on some parts of a document while neglecting other parts. Also, an AI has a finite "token capacity," and can get overwhelmed by details that aren't relevant to the question being asked.
Great article. Good about AI. Far better about thinking. I may be forever in a quandary about active thinking vs passive thinking ( imagining one is thinking due to a vague sense of cognitive presence?).
It would be best if a distinction were made between the different forms of human interaction with AI. There’s “vibe” mode wherein the user uses natural language to communicate with the computer tool. And there’s a programmatic mode, where the user engages AI using programming syntax.
I suspect this article refers specifically to the first mode: natural language. It has to do with contrasting human thought with machine thought. It’s the mode used to write stories, papers, articles, novels, plays, speeches, proposals,theses, reports, etc.
The programming language mode rapidly results in output that when eventually translated into machine language, the only language computers actually can act upon, is bloated, inefficient in execution and difficult to debug. But not so much in every case. The saving grace is AI produces a management result of some kind on the programming task side quickly. Often, this is a short term gain of sorts.
In the vibe, natural language interface mode, AI moves from flattery to apology and back again as one struggles to find anything vaguely resembling thought, much less understanding. Indeed, it’s literally trying to pull hens teeth. Near glacial patience helps. The rewards are vanishingly small.
It’s saving grace is the prose it produces. If you’ve read enough of it, it’s readily identifiable. Verbose by 5 times whats necessary to convey a response. Cadence and structure of sentences, expression and use of digression to appear interesting are programmatically identical.
More than not, AI seems to affirm Pope Leo’s recent assessment.
You had me at the Phaedrus! Thank you for this in-depth discussion. So much great advice that’s helping me think about how to shape AI discussions for the college students I teach.
The comparison with writing, print, calculators, and search makes the AI question feel less binary and more demanding. A tool can weaken one habit while making a different kind of reasoning possible, but only if the user keeps practicing inquiry and evaluation. That distinction between outsourcing memory and building better judgment is the part I found most useful.
Excellent read. I use AI to leverage my time. I use Pocket AI to record my thoughts and ideas for the next chapter in my short stories series. The Watering Hole Chronicles. I can go from a stream of consciousness that I record while doing ranch chores and sometime after dinner I can have a rough draft ready for friends to read.
This is a fantastic and comprehensive article. Thank you for the deep and precise way that you present research that has been framed through the lens of a narrow fear based sensationalism. My writing and research has focused entirely in the last two years on responsible interactive AI use and building methods for users to cultivate a User-Based Governance. I feel urgent that creative, ethical, discerning and expansive humans collectively focus on creating Ai literacy and methods of engagement that empower and reinforce human agency. Your writing does exactly this. I hope to continue to build connections with and contribute to this body of important research and writing.
Thanks! I just testing this out by asking Claude what a beginning knitter might not think about asking before starting, and it gave a great list of things to be cognizant of. It also added that knitting has a steep learning curve and usually takes 10 hours of practice before you feel like you start to understand what you are doing. +1 for Claude.
Then I asked it why I might choose one sort of chimney repair over another. The response sounded very good, but when I went to the links that Claude supposedly used to answer the question, half of them were fake/dead!! -1 for Claude.
So, yes, active use is important, and it seems like right now LLMs are most useful when you already know a little bit about something so you can evaluate the output. Not as useful when you know nothing.
This piece highlights precisely the enormous gap I encounter all the time between intellectually mature and pre-internet educated adults (along with a very few unusual, and precocious, young people) and the vast majority of students in school, K-bacc.
The distinction between active AI use by people who already know how to conduct research, how to ask questions and pursue sustained inquiry, how to evaluate what the tools generate, and especially that asking questions and evaluating responses and results are even things humans do at all, on the one hand, and passive AI use by (currently younger) people who have never learned any of these things and for whom LLMs are simply ATMs with more buttons and steps, makes all the difference.
My older colleagues who love using AI tools and gush about all the new stuff they can do, if they are not also people who teach, especially people who teach students from backgrounds that don't bring with them a culture of reading, or researching, or reasoning, find themselves *completely unable to imagine* what it would be like to use AI e.g. for a writing project without being able to rely on what they already know, on the habits of inquiry and analysis and persistent evaluation developed over a lifetime of upbringing and schooling that forced this development, and offered only isolated shortcuts, like the calculator. They cannot get their heads around what it's like to treat every assignment like a new banking or transit app, just another thing to click on until you get the thing that will get you what you want, namely, in this case, the A. They can't see that without the experience, and the appetite, for the discovery and successful evaluation process, merely adding more steps to the assignment that *mimic* this discovery and evaluation just makes for more clicks that instruct the AI tool to itself *mimic* these processes and preserve them in the output a.k.a. the completed assignment, whether essay, or journal, or project, or poem.
The analogy to the advent of browser-based search engine internet experience actually bears this out, and distinguishes both that advent as well as the new AI tools from the emergence of writing as Plato experienced it, not least because of the staggeringly different timescales: we ourselves are the beneficiaries of a couple of millenia worth of adjustment (in the west) to the losses of oral memory culture and the benefits of writing, benefits that were and are still the subject of laborious evengelicalism in education to reach everyone in the classoom. With the browser- and search engine-internet, we're talking at most 30 years, not even, really, for general exposure and dependency. In my own experience over thirty years of teaching post-secondary students from all kinds of backgrounds I have found that we have--or had--barely begun to make the adjustment the OP here analogizes to the historical response to the advent of writing. That adjustment itself has, in my view, now been completely overrun by the arrival of LLMs and their universal adoption by the vast majority of current students.
When I hear about AI use *for students* that forces, de novo, rather than *relies on*, the development of habits of inquiry and evaluation, as well as actual, non-mimetic, knowledge, I'll be all ears. Until then, I cannot see that this tech is anything but harmful, in general use, for people coming of age intellectually from the turn of this millenium on.
Card Catalog's mission is an outstanding model and guide for thinking about these issues, and for bringing the fruits of our written millenia to an AI world--but it is crucial, in my view, to distinguish in this way between users of AI, because them has to have ears to hear.
Hana, your writing is advanced course work. Solid gold. Thank you so much for this substack. As others have mentioned, I will be referring to it often. And the readers' comments are just as valuable; like sitting in a class with graduate level thinkers. It's worth every penny of a paid subscription.
Outstanding piece Hana, thank you. I will be reviewing this multiple times.
Thank you for posting this information. I attended school, university, and medical school when critical thinking was still embedded in classes. And have been hesitant to use AI or to recommend it to others, having read about "AI brain rot." Now that you have illustrated how it can be used properly, I'm eager to try it.
I wrote a substack a couple of days ago about successful vs unsuccessful AI usage (https://ombreolivier.substack.com/p/ai-mania?r=7yrqz )
In it I used a metaphor I found the basis of somewhere else about AI as a bicycle vs AI as a train. The more I think about it, the more I like this metaphor and it seems consistent with both the scientific evidence you cite and your advice.
AI as bicycle means AI making you more efficient, the way a bike allows you to pedal and travel much faster/further/longer than you can running.
AI as train means you let the AI do the whole job of taking you to the destination for the price of a ticket.
AI as bicycle is good. AI as train is not
When I upload a document, particularly a long document, I often use a prompt like "Please read this document in logical chunks. Create compact internal summaries for each chunk, then synthesize across them so no one section dominates. Focus on key themes, contradictions, caveats, and implications, and especially on [specific question I have about the document]."
This matters because an AI can focus on some parts of a document while neglecting other parts. Also, an AI has a finite "token capacity," and can get overwhelmed by details that aren't relevant to the question being asked.
https://myob371.substack.com/p/my-journey-with-ai?utm_source=share&utm_medium=android&r=770tqq
What a fabulous article, Hana. I can’t wait to include these prompts when I query AI. Thank you again!
I share your articles with so many university colleagues - so incredibly useful, and I really value the non-doom&gloom insights. Thank you.
Great article. Good about AI. Far better about thinking. I may be forever in a quandary about active thinking vs passive thinking ( imagining one is thinking due to a vague sense of cognitive presence?).
It would be best if a distinction were made between the different forms of human interaction with AI. There’s “vibe” mode wherein the user uses natural language to communicate with the computer tool. And there’s a programmatic mode, where the user engages AI using programming syntax.
I suspect this article refers specifically to the first mode: natural language. It has to do with contrasting human thought with machine thought. It’s the mode used to write stories, papers, articles, novels, plays, speeches, proposals,theses, reports, etc.
The programming language mode rapidly results in output that when eventually translated into machine language, the only language computers actually can act upon, is bloated, inefficient in execution and difficult to debug. But not so much in every case. The saving grace is AI produces a management result of some kind on the programming task side quickly. Often, this is a short term gain of sorts.
In the vibe, natural language interface mode, AI moves from flattery to apology and back again as one struggles to find anything vaguely resembling thought, much less understanding. Indeed, it’s literally trying to pull hens teeth. Near glacial patience helps. The rewards are vanishingly small.
It’s saving grace is the prose it produces. If you’ve read enough of it, it’s readily identifiable. Verbose by 5 times whats necessary to convey a response. Cadence and structure of sentences, expression and use of digression to appear interesting are programmatically identical.
More than not, AI seems to affirm Pope Leo’s recent assessment.
Excellent help. And great tips and resources.
You had me at the Phaedrus! Thank you for this in-depth discussion. So much great advice that’s helping me think about how to shape AI discussions for the college students I teach.
The comparison with writing, print, calculators, and search makes the AI question feel less binary and more demanding. A tool can weaken one habit while making a different kind of reasoning possible, but only if the user keeps practicing inquiry and evaluation. That distinction between outsourcing memory and building better judgment is the part I found most useful.
Excellent read. I use AI to leverage my time. I use Pocket AI to record my thoughts and ideas for the next chapter in my short stories series. The Watering Hole Chronicles. I can go from a stream of consciousness that I record while doing ranch chores and sometime after dinner I can have a rough draft ready for friends to read.
This is a fantastic and comprehensive article. Thank you for the deep and precise way that you present research that has been framed through the lens of a narrow fear based sensationalism. My writing and research has focused entirely in the last two years on responsible interactive AI use and building methods for users to cultivate a User-Based Governance. I feel urgent that creative, ethical, discerning and expansive humans collectively focus on creating Ai literacy and methods of engagement that empower and reinforce human agency. Your writing does exactly this. I hope to continue to build connections with and contribute to this body of important research and writing.
Thanks! I just testing this out by asking Claude what a beginning knitter might not think about asking before starting, and it gave a great list of things to be cognizant of. It also added that knitting has a steep learning curve and usually takes 10 hours of practice before you feel like you start to understand what you are doing. +1 for Claude.
Then I asked it why I might choose one sort of chimney repair over another. The response sounded very good, but when I went to the links that Claude supposedly used to answer the question, half of them were fake/dead!! -1 for Claude.
So, yes, active use is important, and it seems like right now LLMs are most useful when you already know a little bit about something so you can evaluate the output. Not as useful when you know nothing.