What are the assumptions about learning embodied in the algorithms?
When I was a junior in high school, my English teacher asked our class to read Hamlet and to write an essay on the line, “There’s a special providence in the fall of a sparrow.” Not surprisingly, given that he was the school’s minister, he believed this line was the “most significant” in the play. Unfortunately, its significance—how, in my teacher’s words, it was “central to making sense of the play”—was lost on me. I wrote a wretched essay, most of which made no sense to me or him. Too bad, because I liked the play. Not surprisingly for me—an angry, cynical teenager at the time—the most significant line was, “I eat the air, promise-crammed.” I could have written a killer essay on that line.
Now, decades later, I am tutoring Lisa, a high school junior whose teacher asked her to write an essay on “the fear of loss and death” in two Dickinson poems. Lisa’s essay was difficult to understand, primarily because she was trying to respond to a prompt that didn’t jibe with her understanding of the two poems; the poems also focused on two very different experiences in the poet’s life, so some readers might not have seen any basis for comparison. Lisa was trying to force connections between an apple and a pot roast. She is a good writer and reader, and, left to her responses, she might have written a fine essay.
English teachers tend to love literature and are often really good at exploring various works and unearthing layers of emotional meaning and insights into a writer’s sense of truth about the human condition. They want to use their understanding to excite their students’ interest in the works and to show them the dazzling rewards of the analytic process. So, their desire to help students appreciate what various authors have created is understandable. A friend and teaching colleague often told me that unless his students could understand A Farewell to Arms as he does, “they simply haven’t read the book.” This perspective is what many English teachers mean when they say they “taught that book.” They presented their sense of the book’s meaning.
When I started teaching, I was no different. My years as a student resulted in my accepting the myth that good English teaching meant providing students with rich interpretations and stimulating prompts that they could spin into erudite essays, proving that they understood literature not only as well as their teacher but in the same way as their teacher. Embedded in my assignments was my interpretive bias: “Analyze the evolution of Frederic Henry’s character throughout the novel, particularly his transition from naïve soldier to disillusioned deserter.” “What about the students who hadn’t reacted to Henry as naïve or disillusioned?” “Explain how the novel portrays the idea of ‘loss of innocence.’” What about the students who see the novel as an exploration of love, whose reaction to “loss of innocence” is, “Huh”? My colleague would simply dismiss these students as incapable of seeing the obvious. My perspective today is that prompts like these capture perfectly writer Ta-Nehisi Coates’s characterization of school as “a jail of other people’s interests.” My insights into and reactions to a book are not much use in achieving the goals I have for my students.
Eventually, I changed my approach. Not only did I grow weary of reading the tortured logic of my students’ writing, but I finally understood what novelists and other artists meant when they said that the reader or audience completed their work: “The book belongs to the reader, not the writer. The reader creates the book.”
Like many English teachers, I will always enjoy the analytic textual search for a writer’s intention and the fun of debating the validity of our differing discoveries with my colleagues. And someday, I hope my students will reach this level of analysis and debate about artists’ intentions. But first, I want them to have a more personal experience—the search for meaning and the discovery of themselves in the text. What matters is not that they find the meaning that I find but that they find their meaning. I want them to feel the excitement of an emotional connection to literature.
(For those of you wondering what any of this has to do with AI, bear with me. We are almost there.)
As a result, I designed my courses like reading clubs. Each of us, students and teacher, selected a work that mattered to us (the only criterion for selection), and everyone in the class read them. The person who selected the work led the discussions about it. Typically, for short works (short stories, poetry, plays, even songs), we spent up to a week (five meetings) on each work, and I recommended that all of us, but especially the discussion leader, read the work three times, two times completely and a third time selectively. To help the students explore the text, I suggested some tools and strategies that they might (or might not) find useful. I offer here an abbreviated version of those that I created for short fiction as illustrations only; I imagine other English teachers could devise better ones:
- First reading:
- Mark anything that stands out, especially repeated ideas, words, images.
- Write any reactions (ideas or feelings) at the point they occur.
- Finish and think, feel, and write random thoughts and impressions.
- Second reading:
- Determine the author’s focus by looking for clues in the following areas:
- title;
- names of characters;
- whose story is being told;
- opening few paragraphs;
- closing paragraph;
- clear symbols;
- central conflict (A vs B) and climax (moment when conflict is resolved—if it is resolved);
- general comments about life (as opposed to lines that simply move the plot, establish the setting, or describe character);
- your understanding of the actions of the characters (based on your own experiences, based on the motivations the writer presents). …
- Determine the author’s focus by looking for clues in the following areas:
- Third reading:
- Determine everything the author is saying about the subject or focus:
- Use all the details, thoughts, reactions, and emotions you collected during the first two readings to establish connections between them and to see if they lead you somewhere;
- move back and forth from feeling to thinking, from the details to the larger picture….
Instead of presenting the results of my analyses—the meaning that I made—I provided a process of analysis that I hoped would lead them to their sense of meaning.
The research of the last 30 years about how people learn resonated with what I experienced in my classrooms, particularly the research of neuroscientist Mary Helen Immordino-Yang (University of Southern California). She and her colleagues offer many insights, three of which I found critical: (1) “Emotion is the rudder for thought”; “we think in the service of emotional goals.” (2) The goal of education is to develop what she calls “transcendent thinking”—an emerging ability in adolescents “to think abstractly and to begin to question who they and others are, based in what they believe in and stand for, including issues impacting themselves, their family and friends, their communities and society in general.” (3) The healthy functioning of three neural networks is essential for meaning-making, moral development, and general wellbeing and happiness—the salience network (that attracts our attention to something that we find emotionally significant); the executive control network (recruited when we are engaged in tasks in the outer world); and the default mode network (used for inner-directed attention and essential for discovering implications and making sense of the world). Developing the ability to toggle between inner focus and outer focus is critical.
As an English teacher, I was particularly struck by Immordino-Yang’s statement, “It is literally neurobiologically impossible to think deeply about things you don’t care about.” My students did not care about my take on books, and my take only succeeded in placing a barrier between them and the books. So they didn’t care about the books, either.
I hoped the changes I adopted in my English classes reflected a larger trend among other teachers. Some teachers abandoned those mini-lecture prompts and encouraged students to choose their essay topics, but not much changed. Now, the grip of that traditional approach to creating writing and discussion prompts seems likely to become more permanent thanks to artificial intelligence. I have written elsewhere about my concern over the insidious belief that AI is simply a useful tool that can help students learn to write, but I have now become even more worried as I read increasingly frequently about “AI teaching assistants” that, according to AI proponents like Tom Daccord, “are appealing and popular, offering to reduce the time teachers spend on one of their most tedious tasks.” Tedious tasks like creating prompts and test questions and grading essays or discussions.
Now the teacher can give the artificial assistant the specifications for the parameters and goals of an assignment; the AI generates the prompt the student will read. Then, AI can have a “discussion” with the student and grade the result. Or the student can write an essay from the prompt, and AI, using some essay grader, can grade the essay and provide “feedback.” Jon Bergmann, another enthusiastic promoter of AI, offers an example of a typical AI-generated prompt in this excerpt:
In George Orwell’s 1984, the oppressive regime of the Party uses fear, surveillance, and propaganda to maintain control over its citizens. The novel explores themes of totalitarianism, the manipulation of truth, and the loss of individuality. … Choose one of the following short stories … and compare how both works explore the themes of control and conformity, and the impact these have on individuals and society. …
The prompt offers a succinct lecture that nicely corrals any maverick student who might decide to think about these works differently. Yet, isn’t thinking differently exactly what we should be encouraging—and, given the research, expect? Aren’t what students discover in a novel or poem and how they discover it the keys to their development as learners and thinkers, the genesis of their sense that books are sources of personal meaning?
While I admit that grading is tedious (and self-defeating), reading and responding to student essays and listening to students discuss literature, though time-consuming and demanding, are two of a teacher’s most essential, hardly “tedious,” obligations. We need to know our students’ developmental and learning progress, trajectories that are revealed only in what they say and write. For example, I read James’s essay, listen to him talk about a poem, and discover he has a very literal mind. When he cites a quotation, he tends to repeat only what it says and ignore its implications for how it supports (or doesn’t) his sense of what the poem means to him. He struggles to articulate the logical connection between the evidence and the claim he makes in a thesis statement. How could I discover this insight into his thinking without reading or hearing his words? How could I create strategies for helping him improve his analytic skills?
Daccord cautions that “AI should not replace teacher judgment,” but teachers cannot form a judgment without engaging with and thinking about what the student has written. Writing our comments as we read their essays is how we sharpen and clarify our thinking about them and, hence, our judgment about what they are thinking. In the AI-assistant model, reading the essays AND THEN having to read and edit the AI comments strike me as a lot of unnecessary extra work that offers nothing in return.
Given my experience with technological assistance over the past many years, I have become aware of how attracted many teachers are to automated shortcuts to some of the essential demands of the job. I recall when teachers began using technology to help them write end-of-term comments. They created banks of clauses and phrases (“always turns in homework on time,” “needs to seek more extra help,” etc.), numbered these, and then just plugged in the numbers and let the computer create the comment. Not surprisingly, the comments grew increasingly thoughtless and homogenized.
I am not anti-technology. I simply believe we need to be careful about deciding when and how we want to use it in our teaching. Efficiency and saving time should not come at the expense of our students’ skills and intellectual development. Shortly after ChatGPT arrived, I heard an interview with a student from Brown University who described how he used AI to generate a topic, a thesis, and an outline for an essay. He said AI saved him hours of work that he would have had to do thinking for himself. Isn’t that thinking the whole point? How do people become meaning-makers without thinking? Oh, wait, of course. They can just adopt the thinking someone else does for them. My English teacher colleague can tell them what to think, or now, AI can tell them what to think. When they become adults, the influencers, pundits, and politicians can tell them what to think.
Perhaps we can be smarter about designing and using technology. Perhaps we can really study and understand the new insights from the sciences of learning, incorporate these into the algorithms, and discover more nourishing roles for AI—for uses by both students and teachers. The one thing we ought not to do is bake outdated and flawed assumptions about learning into the new technology.
About the author
Veteran teacher and administrator Alden S. “Denny” Blodget is the author of “Learning, Schooling and the Brain: New Research vs. Old Assumptions.” He also helped create the Annenberg Foundation’s Neuroscience & The Classroom. He is the editor for TeensParentsTeachers.org, a free online resource focusing on issues affecting young people and the adults who work with them.
