The bigger problem is that a new generation is being trained to rely on "AI" and to regard whatever it delivers as the truth, much as the population at large has been trained to regard the contents of Wikipedia as truth. The Gemini debacle gave us a small glimpse into the political insanity of Google, but few realize that Google's problem is not just in image generation, but in all of the results it returns and suppresses. The frightening thing is what's coming next, which is that Big Tech's Internet censorship regime will declare that their "AI" programs are now the only valid form of "search", and therefore they no longer a need to provide you with old-fashioned search engines. Why sort through your own sources, when the censors at Big Tech can do it for you? READ: Your Coming Brave New World of “Artificial Intelligence”: Your days of having access to search engines are probably numbered: https://daveziffer.substack.com/p/your-coming-brave-new-world-of-artificial
I love reading and feel so sad for those who don’t. There is so much that they bare missing out on.
I hate abridged versions because that seem take out all the uniqueness of a writer’s style and vocabulary. The writing has no “personality” left. The essence that makes the book engaging and unique is gone.
Some value in "executive summaries" -- however generated. But can often lose the sheer delight and deeper understanding of the insights and arguments tabled.
I believe summaries have their place---that is AFTER a thorough reading has been done and used as a review. Do they take the place of actual reading? You know my response. While AI has its place (medical notes, as an 'assist' for students with learning issues, research abstracts), it cannot and must not take the place of a writer's authentic voice. It has little to no place in a classroom (unless used as an assist as I mentioned).
How do you feel about executive summaries of reports or abstracts at the start of research articles? As far back as the 1940s in reading education, teachers were using SQ3R as a study technique. The S for “scan” or “survey” involved getting an overview through scanning of the text before digging in—essentially a quick and dirty summary. In fact, some versions explicitly state to look for a summary and read it first as an advanced organizer to reduce avoidable obstacles to comprehension. The idea of reading a summary after the reading just isn’t supported in mainstream comprehension research. Much of the fears surrounding AI and learning through literacy result from an unwillingness to gain experience using this tool.
I'm certainly not challenging summaries or abstracts written by human beings--I use those all the time myself. But from what I've seen, AI can't be relied on to be accurate at this point. I suppose that's true of some humans too, but I assume that the authors of scientific studies write the abstracts themselves, so those summaries should be trustworthy! In any event, it's hard to imagine any human being making some of the mistakes I saw in the AI-generated workbook of my book.
And I can see how reading a summary before digging into a text would be helpful. What I'm worried about is readers relying on a summary--especially a summary that could be inaccurate or at least oversimplified--INSTEAD of reading the text.
And by the way, teaching students how to summarize a text themselves can be a very effective teaching tool. So I'm not against summaries in general!
I’m relieved to learn that you see benefits to using AI. I’m sorry that I assumed your affiliation with KM meant that you agree with its perspective. I’m very interested in your observation that there is little research on writing. My doctoral program in literacy studies had several courses on composition theory. It’s a rich and robust field. Can you be more specific about what research questions you are trying to answer? My bet is the role of summary in writing instruction is folded into genre theory. For example, creating a paraphrase of a portion of a text to cite in a report is a specific purpose for summarizing—much like creating an executive summary for policy makers or non-expert readers. There are purposes for using a strategy like summary in a writing event that have been studied.
Sentence imitation, sentence combination, sentence rearranging, and sentence expansion are well-researched and indeed are powerful. I can attest to that from my own experience teaching elementary school.
What are your preferred strategies for teaching summarizing? I worked as a reading specialist with a History department in a middle school for two years with a thematic focus on summarizing as a pinnacle reading strategy. As a department we developed a local vocabulary for talking about the relationship between comprehension and summary—summary as a macro strategy for comprehension as an outcome, not summary as an outcome itself—for talking about this theme with students in classrooms. Cross-class fertilization was enriched. We assessed summary writing each trimester across all courses, scored the student-written summary passages, and reflected on for weeks after. AI would be invaluable as a problem solver.
Ask students to read a poem from the 14th century in a collaborative group with a bot. Have one observe and take notes on how the group used the bot. I could go on. Having taught secondary content area reading pedagogy as a literacy education professor for 17 years, I really think it is a mistake for Knowledge Matters to take this perspective on the bot. But it doesn’t surprise me with the evidence I’ve seen of an uneasy courtship involving KM and SoR.
I’ve written very positively about KM on my Substack.
My preferred strategies for teaching summarizing are those that are part of The Writing Revolution method.
I just want to clarify something: You seem to think I'm speaking for "Knowledge Matters." I am not. I don't work for the Knowledge Matters Campaign -- I'm just on the board of their parent organization.
Apologies. The Knowledge Matters group gave me high hopes for sanity to return to literacy pedagogy. I thought you were involved in it as an advocate. I hadn’t considered whether you “work for them.” So you prefer not to be viewed as an advocate for this Campaign? I’ll need to clarify for my readers. Again apologies.
Can you point me in the direction of research vis a vis the Writing Revolution method?
I certainly generally agree with what the Knowledge Matters Campaign advocates for, but I don't speak for them. Nor do they speak for me. Your previous comment indicated that you saw my post as stating a position taken by KMC, but that is not the case. I didn't show my post to anyone there for their approval, nor would I even think of doing something like that, and I have no idea what they might think about the issues I addressed. (And by the way, my post doesn't take a position on AI in general, and it's certainly not a blanket condemnation of it. I have no doubt that there are many benefits to AI when it's used responsibly.)
As for research on The Writing Revolution: The method aligns with much of the research that has been done on writing instruction (there isn't much). That is summarized on the organization's website, here:
Unfortunately there hasn't yet been much experimental research on the method itself (we desperately need more), but I wrote about two studies in a previous post. You can find it here:
I am an educator...never did I use SQ3R...I feel fine about using abstracts at the start of research articles; however, these are not typically part of the K-12 curriculum. My comment was referencing K-12 education and---if I may add, having student WRITE a summary to help them remember what they have read. You will never get me to agree that AI contains a 'writer's voice' no matter how hard you try.
Learning to write good summaries is a complex task which supports memory for content and improves comprehension processes as well. Don’t you think high school students should have experiences reading scientific articles? AI could be useful in such work. I agree with you totally AI has no ‘voice’ at all. AI cannot read nor can it write. It processes information and has no more understanding of what it outputs than a Google search. What did I write to lead you to the conclusion that I want to agree with me that AI has a voice? Btw, SQ3R is the ancestor of many early content area reading lesson templates like DR-TA, KWL, PReP, Anticipation Guides, and many more. The most recent research that changes its paradigm a bit comes from Patricia Alexander and the theory of relational reasoning. I have a post on it.
Nope. Neither Faulkner nor Hemingway. I agree totally. When it comes to interpretation, AI sucks. When you need information to get a foothold in what the woods even mean, AI helps. Well said! I am in awe of the power of your writing here. You model your message. Thanks so much! Keep in touch. You and I are simpatico
I just found you through Racket. Thank you! Your substack is exactly what I’ve been looking for without knowing it.
I can’t imagine AI having a constructive role in education yet, if ever. I struggled to learn everything I ever learned. The struggle is worth it to me because I know that the Vulcan Mind Meld is fiction. Not enough people have a grip on reality, or even want to, and that is a huge problem.
As someone who faced challenges learning to read, AI has been a godsend. Used appropriately, AI has expanded the market of readers who can access once indecipherable passages of text into something meaningful. AI has helped me to become a better reader and writer. My struggles with language have been shepherded to a more refined understanding, not less. AI is a great remedial tool to lift up the struggling to get to a better place in their reading and writing. Of course, AI is limiting to those who don’t truly need assistance, just like braces would limit those not suffering from a muscle disease.
Wow. I worked in early computing in the late 70’s early 80’s and we sometimes discussed how AI could really damage our culture. I can’t imagine the frustration you must feel about this. Anything to “game the system” and make a buck. I’m not a fan of workbooks, but in this case, I’d say go for it.
Thank you for telling this to the world, j.e. This sentence is beautiful: “AI has expanded the market of readers who can access once indecipherable passages of text into something meaningful.” I’m not dyslexic, but I often encounter texts that are indecipherable, and AI is a godsend to me as well. It doesn’t help much with authors like James Joyce, Clarice Lispector, or William Faulkner, but it brings previously daunting texts within reach. Lots of people make hasty generalizations from one instance with a crazy bot. If Natalie had used the bot to create a workbook herself, providing the bot with clear goals, she wouldn’t be so disenthralled with it. I see this all the time among academics. Use it first. Then criticize it. People were once afraid of airplanes.
I don’t see AI as a brace so much as a semiotic problem solving tool. For example, I am among the unfortunates who never read Homer. I tried several times, but I couldn’t make heads or tails of the Iliad, which I tried first. After reading the Iliad, the Odyssey was much less indecipherable. I read the Iliad with AI every step of the way to help me get past the hundreds of gods, goddesses, and soldiers that interact among Hector, Achilles, Agamemnon, Nestor, and Paris. It took me three months.
Don’t get me wrong. I’m not an expert on Homer by any means. But I can now read and talk with experts. AI could never have read Homer for me. But I honestly don’t know how I could have read it apart from a graduate seminar without AI.
Fears about AI doing the reading are overstated at best. AI cannot read nor can it write. But the only way to learn that is to use AI regularly
I am not an educator or a writer. I am a pharmacist who grew up poor on a farm. Reading and learning were greatly valued by my parents and extended family. We lived in a very rural area so cable and satellite were not affordable or available options. We had the big 3, as well as PBS. We spent much of our time reading physical books.
My house is full of books. I have hundreds I have read more than once, and just as many more I have yet to read. I read fiction, nonfiction, and a little of almost any category. There are so many stories and poems in my mind that I can pull forward to occupy or distract myself. I have read Skippyjon Jones to my nieces so many times, I can still remember and recite the first few lines. Or I think about a new concept I have learned or a different perspective in viewing history or events or interrelationships.
Using summaries can be very helpful in researching an answer to a question a patient has about a medication or health condition. It helps me quickly decide whether the article contains the information I am searching for. Scientific writing is very different from literature, biography, and other types of writing where the nuances of the words one chooses makes a difference in meaning. Allowing AI to use a thesaurus to pick words similar to but not the same as the ones the writer chose can greatly change what one takes away from what one has read.
I have an Oxford comma and to me, there is a difference in meaning when it is used versus when it is not. (Thank you Eats, Shoots and Leaves).
I love the English language and the half million words in it. I love being able to express myself with nuance for the exact meaning I intend. I would be so limited in French or other languages with only fifty thousand words or so.
(Sorry for rambling and getting off topic—I had interruptions)
AI can have advantages for some in certain situations. I would hate to have it take over and replace human thought and imagination and creativity. It can mimic but can it truly create? It is limited by the parameters and information it is provided with. Would AI be able to produce the writing of William Faulkner or Ernest Hemingway without first having been provided with examples of their writing?
There’s nothing inherently wrong with reading a summary of a text, fiction or nonfiction. For example, research articles include an abstract to provide a basis for deciding whether the text is relevant to the research problem or question. Using a bot to generate a summary for this purpose makes sense as well. In many public school classrooms, students don’t have much opportunity to do this sort of literacy work; so they don’t accumulate experience with summaries as a filter to produce a relevant reading list. When learners get a steady diet of assigned reading with externally prescribed purposes without regard to their invested work in projects, a summary is a short version of a long text.
The bot is a game changer when it comes to close reading. Readers need much more instruction in technical issues of situated meanings of words wherein syntax and phrase structure rules create nuances. Shifting from a dictionaries definition to an articulated situated meaning involves critical thinking while engaging in reading between the lines. The bot is an incredibly powerful tool.
It takes a great deal of experience using the bot during reading to discover its capacity to peel back layers of context. As a PhD in language and literacy with an MA in English, having used the bot daily since Nov 2022 when OpenAI released GPT, I’ve found myself gradually moving from resistance and fear to optimism. But teachers will need to change their emphasis on instructional control. When schoolwork is assigned piece by piece for someone else’s purpose and kids are staring at quizzes and tests and grades, the bot doesn’t fit very well.
The bigger problem is that a new generation is being trained to rely on "AI" and to regard whatever it delivers as the truth, much as the population at large has been trained to regard the contents of Wikipedia as truth. The Gemini debacle gave us a small glimpse into the political insanity of Google, but few realize that Google's problem is not just in image generation, but in all of the results it returns and suppresses. The frightening thing is what's coming next, which is that Big Tech's Internet censorship regime will declare that their "AI" programs are now the only valid form of "search", and therefore they no longer a need to provide you with old-fashioned search engines. Why sort through your own sources, when the censors at Big Tech can do it for you? READ: Your Coming Brave New World of “Artificial Intelligence”: Your days of having access to search engines are probably numbered: https://daveziffer.substack.com/p/your-coming-brave-new-world-of-artificial
For more insight into Google's manic politics, also READ: Three Companies Now Run The World: https://daveziffer.substack.com/p/three-companies-now-run-the-world
I love reading and feel so sad for those who don’t. There is so much that they bare missing out on.
I hate abridged versions because that seem take out all the uniqueness of a writer’s style and vocabulary. The writing has no “personality” left. The essence that makes the book engaging and unique is gone.
Some value in "executive summaries" -- however generated. But can often lose the sheer delight and deeper understanding of the insights and arguments tabled.
I believe summaries have their place---that is AFTER a thorough reading has been done and used as a review. Do they take the place of actual reading? You know my response. While AI has its place (medical notes, as an 'assist' for students with learning issues, research abstracts), it cannot and must not take the place of a writer's authentic voice. It has little to no place in a classroom (unless used as an assist as I mentioned).
How do you feel about executive summaries of reports or abstracts at the start of research articles? As far back as the 1940s in reading education, teachers were using SQ3R as a study technique. The S for “scan” or “survey” involved getting an overview through scanning of the text before digging in—essentially a quick and dirty summary. In fact, some versions explicitly state to look for a summary and read it first as an advanced organizer to reduce avoidable obstacles to comprehension. The idea of reading a summary after the reading just isn’t supported in mainstream comprehension research. Much of the fears surrounding AI and learning through literacy result from an unwillingness to gain experience using this tool.
I'm certainly not challenging summaries or abstracts written by human beings--I use those all the time myself. But from what I've seen, AI can't be relied on to be accurate at this point. I suppose that's true of some humans too, but I assume that the authors of scientific studies write the abstracts themselves, so those summaries should be trustworthy! In any event, it's hard to imagine any human being making some of the mistakes I saw in the AI-generated workbook of my book.
And I can see how reading a summary before digging into a text would be helpful. What I'm worried about is readers relying on a summary--especially a summary that could be inaccurate or at least oversimplified--INSTEAD of reading the text.
And by the way, teaching students how to summarize a text themselves can be a very effective teaching tool. So I'm not against summaries in general!
I’m relieved to learn that you see benefits to using AI. I’m sorry that I assumed your affiliation with KM meant that you agree with its perspective. I’m very interested in your observation that there is little research on writing. My doctoral program in literacy studies had several courses on composition theory. It’s a rich and robust field. Can you be more specific about what research questions you are trying to answer? My bet is the role of summary in writing instruction is folded into genre theory. For example, creating a paraphrase of a portion of a text to cite in a report is a specific purpose for summarizing—much like creating an executive summary for policy makers or non-expert readers. There are purposes for using a strategy like summary in a writing event that have been studied.
Sentence imitation, sentence combination, sentence rearranging, and sentence expansion are well-researched and indeed are powerful. I can attest to that from my own experience teaching elementary school.
What are your preferred strategies for teaching summarizing? I worked as a reading specialist with a History department in a middle school for two years with a thematic focus on summarizing as a pinnacle reading strategy. As a department we developed a local vocabulary for talking about the relationship between comprehension and summary—summary as a macro strategy for comprehension as an outcome, not summary as an outcome itself—for talking about this theme with students in classrooms. Cross-class fertilization was enriched. We assessed summary writing each trimester across all courses, scored the student-written summary passages, and reflected on for weeks after. AI would be invaluable as a problem solver.
Ask students to read a poem from the 14th century in a collaborative group with a bot. Have one observe and take notes on how the group used the bot. I could go on. Having taught secondary content area reading pedagogy as a literacy education professor for 17 years, I really think it is a mistake for Knowledge Matters to take this perspective on the bot. But it doesn’t surprise me with the evidence I’ve seen of an uneasy courtship involving KM and SoR.
I’ve written very positively about KM on my Substack.
My preferred strategies for teaching summarizing are those that are part of The Writing Revolution method.
I just want to clarify something: You seem to think I'm speaking for "Knowledge Matters." I am not. I don't work for the Knowledge Matters Campaign -- I'm just on the board of their parent organization.
Apologies. The Knowledge Matters group gave me high hopes for sanity to return to literacy pedagogy. I thought you were involved in it as an advocate. I hadn’t considered whether you “work for them.” So you prefer not to be viewed as an advocate for this Campaign? I’ll need to clarify for my readers. Again apologies.
Can you point me in the direction of research vis a vis the Writing Revolution method?
I certainly generally agree with what the Knowledge Matters Campaign advocates for, but I don't speak for them. Nor do they speak for me. Your previous comment indicated that you saw my post as stating a position taken by KMC, but that is not the case. I didn't show my post to anyone there for their approval, nor would I even think of doing something like that, and I have no idea what they might think about the issues I addressed. (And by the way, my post doesn't take a position on AI in general, and it's certainly not a blanket condemnation of it. I have no doubt that there are many benefits to AI when it's used responsibly.)
As for research on The Writing Revolution: The method aligns with much of the research that has been done on writing instruction (there isn't much). That is summarized on the organization's website, here:
https://www.thewritingrevolution.org/method/research/
Unfortunately there hasn't yet been much experimental research on the method itself (we desperately need more), but I wrote about two studies in a previous post. You can find it here:
https://nataliewexler.substack.com/p/to-improve-students-writing-teach
I am an educator...never did I use SQ3R...I feel fine about using abstracts at the start of research articles; however, these are not typically part of the K-12 curriculum. My comment was referencing K-12 education and---if I may add, having student WRITE a summary to help them remember what they have read. You will never get me to agree that AI contains a 'writer's voice' no matter how hard you try.
Learning to write good summaries is a complex task which supports memory for content and improves comprehension processes as well. Don’t you think high school students should have experiences reading scientific articles? AI could be useful in such work. I agree with you totally AI has no ‘voice’ at all. AI cannot read nor can it write. It processes information and has no more understanding of what it outputs than a Google search. What did I write to lead you to the conclusion that I want to agree with me that AI has a voice? Btw, SQ3R is the ancestor of many early content area reading lesson templates like DR-TA, KWL, PReP, Anticipation Guides, and many more. The most recent research that changes its paradigm a bit comes from Patricia Alexander and the theory of relational reasoning. I have a post on it.
Nope. Neither Faulkner nor Hemingway. I agree totally. When it comes to interpretation, AI sucks. When you need information to get a foothold in what the woods even mean, AI helps. Well said! I am in awe of the power of your writing here. You model your message. Thanks so much! Keep in touch. You and I are simpatico
I just found you through Racket. Thank you! Your substack is exactly what I’ve been looking for without knowing it.
I can’t imagine AI having a constructive role in education yet, if ever. I struggled to learn everything I ever learned. The struggle is worth it to me because I know that the Vulcan Mind Meld is fiction. Not enough people have a grip on reality, or even want to, and that is a huge problem.
As someone who faced challenges learning to read, AI has been a godsend. Used appropriately, AI has expanded the market of readers who can access once indecipherable passages of text into something meaningful. AI has helped me to become a better reader and writer. My struggles with language have been shepherded to a more refined understanding, not less. AI is a great remedial tool to lift up the struggling to get to a better place in their reading and writing. Of course, AI is limiting to those who don’t truly need assistance, just like braces would limit those not suffering from a muscle disease.
Wow. I worked in early computing in the late 70’s early 80’s and we sometimes discussed how AI could really damage our culture. I can’t imagine the frustration you must feel about this. Anything to “game the system” and make a buck. I’m not a fan of workbooks, but in this case, I’d say go for it.
Thank you for telling this to the world, j.e. This sentence is beautiful: “AI has expanded the market of readers who can access once indecipherable passages of text into something meaningful.” I’m not dyslexic, but I often encounter texts that are indecipherable, and AI is a godsend to me as well. It doesn’t help much with authors like James Joyce, Clarice Lispector, or William Faulkner, but it brings previously daunting texts within reach. Lots of people make hasty generalizations from one instance with a crazy bot. If Natalie had used the bot to create a workbook herself, providing the bot with clear goals, she wouldn’t be so disenthralled with it. I see this all the time among academics. Use it first. Then criticize it. People were once afraid of airplanes.
I don’t see AI as a brace so much as a semiotic problem solving tool. For example, I am among the unfortunates who never read Homer. I tried several times, but I couldn’t make heads or tails of the Iliad, which I tried first. After reading the Iliad, the Odyssey was much less indecipherable. I read the Iliad with AI every step of the way to help me get past the hundreds of gods, goddesses, and soldiers that interact among Hector, Achilles, Agamemnon, Nestor, and Paris. It took me three months.
Don’t get me wrong. I’m not an expert on Homer by any means. But I can now read and talk with experts. AI could never have read Homer for me. But I honestly don’t know how I could have read it apart from a graduate seminar without AI.
Fears about AI doing the reading are overstated at best. AI cannot read nor can it write. But the only way to learn that is to use AI regularly
I am not an educator or a writer. I am a pharmacist who grew up poor on a farm. Reading and learning were greatly valued by my parents and extended family. We lived in a very rural area so cable and satellite were not affordable or available options. We had the big 3, as well as PBS. We spent much of our time reading physical books.
My house is full of books. I have hundreds I have read more than once, and just as many more I have yet to read. I read fiction, nonfiction, and a little of almost any category. There are so many stories and poems in my mind that I can pull forward to occupy or distract myself. I have read Skippyjon Jones to my nieces so many times, I can still remember and recite the first few lines. Or I think about a new concept I have learned or a different perspective in viewing history or events or interrelationships.
Using summaries can be very helpful in researching an answer to a question a patient has about a medication or health condition. It helps me quickly decide whether the article contains the information I am searching for. Scientific writing is very different from literature, biography, and other types of writing where the nuances of the words one chooses makes a difference in meaning. Allowing AI to use a thesaurus to pick words similar to but not the same as the ones the writer chose can greatly change what one takes away from what one has read.
I have an Oxford comma and to me, there is a difference in meaning when it is used versus when it is not. (Thank you Eats, Shoots and Leaves).
I love the English language and the half million words in it. I love being able to express myself with nuance for the exact meaning I intend. I would be so limited in French or other languages with only fifty thousand words or so.
(Sorry for rambling and getting off topic—I had interruptions)
AI can have advantages for some in certain situations. I would hate to have it take over and replace human thought and imagination and creativity. It can mimic but can it truly create? It is limited by the parameters and information it is provided with. Would AI be able to produce the writing of William Faulkner or Ernest Hemingway without first having been provided with examples of their writing?
There’s nothing inherently wrong with reading a summary of a text, fiction or nonfiction. For example, research articles include an abstract to provide a basis for deciding whether the text is relevant to the research problem or question. Using a bot to generate a summary for this purpose makes sense as well. In many public school classrooms, students don’t have much opportunity to do this sort of literacy work; so they don’t accumulate experience with summaries as a filter to produce a relevant reading list. When learners get a steady diet of assigned reading with externally prescribed purposes without regard to their invested work in projects, a summary is a short version of a long text.
The bot is a game changer when it comes to close reading. Readers need much more instruction in technical issues of situated meanings of words wherein syntax and phrase structure rules create nuances. Shifting from a dictionaries definition to an articulated situated meaning involves critical thinking while engaging in reading between the lines. The bot is an incredibly powerful tool.
It takes a great deal of experience using the bot during reading to discover its capacity to peel back layers of context. As a PhD in language and literacy with an MA in English, having used the bot daily since Nov 2022 when OpenAI released GPT, I’ve found myself gradually moving from resistance and fear to optimism. But teachers will need to change their emphasis on instructional control. When schoolwork is assigned piece by piece for someone else’s purpose and kids are staring at quizzes and tests and grades, the bot doesn’t fit very well.
Thanks, Natalie!