A fascinating piece by Steven Johnson. I am a senior (89) who takes an occipital course at NC State University's OLLI (Osher Lifelong Learning Institute). They are devoting a major part of this next semester to AI. I wish Steven could lecture there on AI and Notebook LM.
Full disclosure. Steven is my nephew and my son's best friend. They attended high school (St Albans) and college (Brown) together. I learn from them every day. Nothing artificial about this.
Very interesting. I've been very curious about this too. Recently, I worked with Claude to design a set of parameters for it called "learning mode." When I tell it to engage "learning mode," it has to engage with me according to a set of rules designed to minimize cognitive offloading (and I suppose maximize what you call cognitive uploading). It means it sometimes asks me annoying questions like, "Well, what do you think the passage means?" instead of summarizing it for me — but it's been a fun constraint. I'm planning to continue testing it.
That was the only thing I was tempted to add to your bullet point list: A kind of conversational back and forth where AI checks your comprehension of the sources you've read, and helps make sure you understand everything correctly.
Glad to hear people are working on this sort of thing!
Working with AI is like working on a group project. If your idea of doing a group project is to let someone else do all the work, then of course you will get nothing out of it. But if everyone contributes, then it can be a positive experience.
Lewis Mumford argued that the difference between a machine and a tool is that the machine dictates when and how it should be used, whereas a tool lets the user decide.
AI is here to stay, the question is whether we will become cogs in a machine or master it as a craftsperson masters their tools. This is one of the most practical essays I've read on how to do that, with a clear example in the context of research and writing. Thanks!
Much needed nuance in how the relation between AI use and thinking can work. Reference to NotebookLM make a lot of sense in this context. It is an excellent research and thinking partner.
> So part of the conviction that AI is in a zero-sum battle with human thinking stems from people simply not following the progress since 2022, both in the underlying models themselves and the application frameworks we have built around them.
One thing I love about this piece is you actually gave us clear examples of how we could use AI to enhance our thinking and learning, and not replace our thinking.
As a university teacher and researcher, I don’t think that’s the biggest problem we face when it comes to AI. Instead, we must ask: Is this technology operated ethically? Has it paid the authors for the data and information on which it relies? Is it safe and does it respect our freedoms and privacy? Does it operate sustainably, with respect for the environment? Does it pay fair taxes on its profits? Etc., etc.
Or, what if we humans are the editors and the AI is the writer? It's hard to be a good editor; even harder to be a good editor-in-chief—at least as hard as being a good writer. Imagine yourself with the power to decide _what_ articles go into the New Yorker or The Economist and _which_ writers will do each piece -- that's what AI lets you do.
I have been using NLM for more than a year now - also Claude and Perplexity AI for a project I call, "Antichaos." It takes the form of a new Substack, "Antichaos" at resonantrules.substack … , and in the future, a book, and maybe a university course. The AI capabilities from multiple sources form integrated toolset. Without them I could not have made progress on this multidisciplinary project to understand the causes, impacts, and mitigation mechanisms related to the current chaotic environment in information, politics, and of course, climate. I look at this in relation to an ethical principle I call Future Agency, the ability of humans and collectives to predict and mold the future, what has been called Pursuit of Happiness, with maybe a little more specificity and rigor.
Thank you Steven Johnson for moving the AI ball forward. Far from limiting creativity, the AI tools allow, and maybe, force deeper thinking and the ability to see connections that were previously invisible.
Multiple fields are changing rapidly as a result of the new R&D machines: law, coding, scientific investigation, and the humanities as well. Standing still or banning access to the new environment is not an option. Research that had taken months of work now takes a few hours, with focus and progress that is more intense than I have experienced before. I'm sure that in time I will discover unanticipated and valuable ways of using the tools. And remarkably, research and writing, while still work, have become something closer to joy than struggle.
Wow! Insightful piece. And I had lots of thoughts as I read through your article. Too many to recount here. But many of them pointed out assumptions in your arguments that, if they were not true, (and I cynically believed that most of those that I noted would not be true), would then result in a problem with your suggestions.
However, one way to address the issue might be to use AI to help us revamp how we evaluate the students' comprehension and grasp of the materials. I think this is what needs to change in order to accommodate this shift in technology and better build in motivation for the students to do the hard work of making sure they gain comprehension. By not using old methods of testing comprehension and mastery that were built before the use of AI was an option.
Quite frankly, I think the emergence of AI, while it decimates many of the ways that we have done things in the past, will ultimately be helpful... for we will often find that the solution for the decimation takes us to a place that is better than the methods that we used before AI.
For example, tests, papers, multiple-choice exams, A through F grading systems were never the best way of testing or measuring comprehension or mastery. AI might be inviting us to come higher and, ultimately, motivate us and help us to produce better methods if we're smart about how we approach the challenge before us and how we use the tool.
P.S. - We also need to figure out how to take into account the difference in comprehension when we are actually reading on paper and writing ourselves. Not sure how to fit this all in but it is worth considering.
Agreed - we can read something and have a “surface level” understanding - and whether that’s “deep” enough to write about it = I think that’s a WHOLE new level of comprehension?
Also - those terms I just used (surface & deep?) - I don’t really like them much?
I just don’t have “alternative” terms that I like better - and they are used to mean SO MANY different things in research???
I like Isobel Beck’s research on learning word meanings with different levels of understanding as it’s more concrete … Hope that makes sense? 👍
Really liked your approach to how to use AI toward the end. You were basically describing much of the role my advisor played when I was writing my dissertation at Purdue(30 years ago!). She was a great advisor (Susan Curtis), and most would not have done this good a job nor as good a job as what you describe.
Great article, with relevant ways to use ai to upload more cognition, cognition uploading is a great term to describe aigent assisted learning.
For younger learners, The Young Lady’s Illustrated Primer from the sci-fi book The Diamond Age remains an excellent goal.
I recall working at IBM and advising them to use Watson as an ai tutor for English (as a Second Language). This remains in my mind as the most likely first profitable subject for teaching, and so far I haven’t heard of any breakout killer app. Duo Lingo ain’t it, tho it’s a fine early step for getting some vocabulary & a little grammar.
Not sure if Google offers courses to go from B1 B2 B3 C1 C2, with testing, but they should.
Also - your new term “cognitive uploading” is distinct from the other TWO terms of “cognitive offloading” & “cognitive outsourcing” = THREE Key concepts that students & teachers (& maybe workers & businesses) need to learn & understand DEEPLY!
AND - for your information & others - I learn SO MUCH from reading the comments in every Substack (& especially yours now) - and I can’t keep up with the pace of what’s posted!
NOW - I’m slowing down & allowing time to think through how new learning “fits” with what I already know - if that makes sense?
I’ve just liked many of your comments from others here & that’s ANOTHER complement to you - as you are TRULY a leader in this Substack & Technology world that’s still developing!
This is the first post of yours I’ve read - thanks to Michael Wagner who referenced you in his recent post!!
I have learned SO MUCH from this ONE post and so I’ve subscribed at the FREE level - mainly because I’m a part-retired academic & don’t have a lot of spare cash…
Your points really resonate with me as this is EXACTLY the reason I’ve been reading about AI on Substack for MONTHS & the same direction I wanted to write an intervention program for schools - starting in elementary (primary) grades.
AI is NEVER going away - so - rather than having young people “opting out” of learning, I believe they need to learn WITH and USING AI.
I’ve been searching for how to do this & you’ve helped my thinking in terms of direction as well as WONDERFUL practical examples I can now model for teachers & students.
MANY THANKS again & I will be looking at how to use Your NotebookLM as a way forward!
Much appreciated & keep posting - so I & others can keep learning!! 👍👍♥️♥️
A fascinating piece by Steven Johnson. I am a senior (89) who takes an occipital course at NC State University's OLLI (Osher Lifelong Learning Institute). They are devoting a major part of this next semester to AI. I wish Steven could lecture there on AI and Notebook LM.
Full disclosure. Steven is my nephew and my son's best friend. They attended high school (St Albans) and college (Brown) together. I learn from them every day. Nothing artificial about this.
AGREED - I’m a “bit older” too & just learned HEAPS” from this one post! 👍👍
Very interesting. I've been very curious about this too. Recently, I worked with Claude to design a set of parameters for it called "learning mode." When I tell it to engage "learning mode," it has to engage with me according to a set of rules designed to minimize cognitive offloading (and I suppose maximize what you call cognitive uploading). It means it sometimes asks me annoying questions like, "Well, what do you think the passage means?" instead of summarizing it for me — but it's been a fun constraint. I'm planning to continue testing it.
That was the only thing I was tempted to add to your bullet point list: A kind of conversational back and forth where AI checks your comprehension of the sources you've read, and helps make sure you understand everything correctly.
Glad to hear people are working on this sort of thing!
Working with AI is like working on a group project. If your idea of doing a group project is to let someone else do all the work, then of course you will get nothing out of it. But if everyone contributes, then it can be a positive experience.
Lewis Mumford argued that the difference between a machine and a tool is that the machine dictates when and how it should be used, whereas a tool lets the user decide.
AI is here to stay, the question is whether we will become cogs in a machine or master it as a craftsperson masters their tools. This is one of the most practical essays I've read on how to do that, with a clear example in the context of research and writing. Thanks!
Much needed nuance in how the relation between AI use and thinking can work. Reference to NotebookLM make a lot of sense in this context. It is an excellent research and thinking partner.
> So part of the conviction that AI is in a zero-sum battle with human thinking stems from people simply not following the progress since 2022, both in the underlying models themselves and the application frameworks we have built around them.
One thing I love about this piece is you actually gave us clear examples of how we could use AI to enhance our thinking and learning, and not replace our thinking.
As a university teacher and researcher, I don’t think that’s the biggest problem we face when it comes to AI. Instead, we must ask: Is this technology operated ethically? Has it paid the authors for the data and information on which it relies? Is it safe and does it respect our freedoms and privacy? Does it operate sustainably, with respect for the environment? Does it pay fair taxes on its profits? Etc., etc.
Or, what if we humans are the editors and the AI is the writer? It's hard to be a good editor; even harder to be a good editor-in-chief—at least as hard as being a good writer. Imagine yourself with the power to decide _what_ articles go into the New Yorker or The Economist and _which_ writers will do each piece -- that's what AI lets you do.
I have been using NLM for more than a year now - also Claude and Perplexity AI for a project I call, "Antichaos." It takes the form of a new Substack, "Antichaos" at resonantrules.substack … , and in the future, a book, and maybe a university course. The AI capabilities from multiple sources form integrated toolset. Without them I could not have made progress on this multidisciplinary project to understand the causes, impacts, and mitigation mechanisms related to the current chaotic environment in information, politics, and of course, climate. I look at this in relation to an ethical principle I call Future Agency, the ability of humans and collectives to predict and mold the future, what has been called Pursuit of Happiness, with maybe a little more specificity and rigor.
Thank you Steven Johnson for moving the AI ball forward. Far from limiting creativity, the AI tools allow, and maybe, force deeper thinking and the ability to see connections that were previously invisible.
Multiple fields are changing rapidly as a result of the new R&D machines: law, coding, scientific investigation, and the humanities as well. Standing still or banning access to the new environment is not an option. Research that had taken months of work now takes a few hours, with focus and progress that is more intense than I have experienced before. I'm sure that in time I will discover unanticipated and valuable ways of using the tools. And remarkably, research and writing, while still work, have become something closer to joy than struggle.
Wow! Insightful piece. And I had lots of thoughts as I read through your article. Too many to recount here. But many of them pointed out assumptions in your arguments that, if they were not true, (and I cynically believed that most of those that I noted would not be true), would then result in a problem with your suggestions.
However, one way to address the issue might be to use AI to help us revamp how we evaluate the students' comprehension and grasp of the materials. I think this is what needs to change in order to accommodate this shift in technology and better build in motivation for the students to do the hard work of making sure they gain comprehension. By not using old methods of testing comprehension and mastery that were built before the use of AI was an option.
Quite frankly, I think the emergence of AI, while it decimates many of the ways that we have done things in the past, will ultimately be helpful... for we will often find that the solution for the decimation takes us to a place that is better than the methods that we used before AI.
For example, tests, papers, multiple-choice exams, A through F grading systems were never the best way of testing or measuring comprehension or mastery. AI might be inviting us to come higher and, ultimately, motivate us and help us to produce better methods if we're smart about how we approach the challenge before us and how we use the tool.
P.S. - We also need to figure out how to take into account the difference in comprehension when we are actually reading on paper and writing ourselves. Not sure how to fit this all in but it is worth considering.
Agreed - we can read something and have a “surface level” understanding - and whether that’s “deep” enough to write about it = I think that’s a WHOLE new level of comprehension?
Also - those terms I just used (surface & deep?) - I don’t really like them much?
I just don’t have “alternative” terms that I like better - and they are used to mean SO MANY different things in research???
I like Isobel Beck’s research on learning word meanings with different levels of understanding as it’s more concrete … Hope that makes sense? 👍
Really liked your approach to how to use AI toward the end. You were basically describing much of the role my advisor played when I was writing my dissertation at Purdue(30 years ago!). She was a great advisor (Susan Curtis), and most would not have done this good a job nor as good a job as what you describe.
Great article, with relevant ways to use ai to upload more cognition, cognition uploading is a great term to describe aigent assisted learning.
For younger learners, The Young Lady’s Illustrated Primer from the sci-fi book The Diamond Age remains an excellent goal.
I recall working at IBM and advising them to use Watson as an ai tutor for English (as a Second Language). This remains in my mind as the most likely first profitable subject for teaching, and so far I haven’t heard of any breakout killer app. Duo Lingo ain’t it, tho it’s a fine early step for getting some vocabulary & a little grammar.
Not sure if Google offers courses to go from B1 B2 B3 C1 C2, with testing, but they should.
Also - your new term “cognitive uploading” is distinct from the other TWO terms of “cognitive offloading” & “cognitive outsourcing” = THREE Key concepts that students & teachers (& maybe workers & businesses) need to learn & understand DEEPLY!
AND - for your information & others - I learn SO MUCH from reading the comments in every Substack (& especially yours now) - and I can’t keep up with the pace of what’s posted!
NOW - I’m slowing down & allowing time to think through how new learning “fits” with what I already know - if that makes sense?
I’ve just liked many of your comments from others here & that’s ANOTHER complement to you - as you are TRULY a leader in this Substack & Technology world that’s still developing!
MANY THANKS! 👍👍♥️
This is the first post of yours I’ve read - thanks to Michael Wagner who referenced you in his recent post!!
I have learned SO MUCH from this ONE post and so I’ve subscribed at the FREE level - mainly because I’m a part-retired academic & don’t have a lot of spare cash…
Your points really resonate with me as this is EXACTLY the reason I’ve been reading about AI on Substack for MONTHS & the same direction I wanted to write an intervention program for schools - starting in elementary (primary) grades.
AI is NEVER going away - so - rather than having young people “opting out” of learning, I believe they need to learn WITH and USING AI.
I’ve been searching for how to do this & you’ve helped my thinking in terms of direction as well as WONDERFUL practical examples I can now model for teachers & students.
MANY THANKS again & I will be looking at how to use Your NotebookLM as a way forward!
Much appreciated & keep posting - so I & others can keep learning!! 👍👍♥️♥️