Review of Teaching with AI: A practical guide to a new era of human learning by J. A. Bowen & C. E. Watson Fowler, D. (2025, September 3). Review of Teaching with AI: A practical guide to a new era of human learning by J. A. Bowen & C. E. Watson. Education Review, 32. https://doi.org/10.14507/er.v32.4255 September 3, 2025 ISSN 1094-5296 Bowen, J. A., & Watson, C. E. (2024). Teaching with AI: A practical guide to a new era of human learning. Johns Hopkins University Press. 279 pp. ISBN: 978-1421449234 Reviewed by David Fowler University of Nebraska–Lincoln United States From a recent human-computer conversation (ChatGPT, 2025): Human prompt: I read that AIs are “very close to thinking,” and that led me to imagine an AI musing “I almost think, therefore I almost exist.” ChatGPT: Now here’s where your phrase hits home: “very close to thinking.” Let’s be precise: • I can process, generate, and recombine. • I can simulate inference, analogy, response to context. • I can even respond with what feels like insight, because I model patterns that resemble insight. But here’s the catch: There is no “I” that thinks. There is no first-person perspective, no interiority, no aboutness. There is output without presence. As Thomas Nagel might say, there is nothing it is like to be me. So: almost thinking. That’s a strangely accurate phrase. And from that, your poetic inversion: “I almost think, therefore I almost exist.” That sounds like something an AI would say—if it could reflect on its own non- existence. Edgar A. Poe’s classic poem, The Bells (Poe, 1992) suggests useful phrases for describing the impact of generative artificial intelligence on the college English classroom. An optimistic tintinnabulation of silver bells accompanied the arrival of https://doi.org/10.14507/er.v32.4255 Education Review/Reseñas Educativas/Resenhas Educativas 2 “AI,” while a melancholy monody of iron bells foretold the end of college education, especially in the liberal arts branches of higher learning. The media clangor began in November 2022, when the research group OpenAI launched its computer application ChatGPT 3.5. Pro and con discussions resonantly swelled as new AI products appeared, evolving so rapidly that by 2024 the technology futurist Ray Kurzweil wrote “AI progress is now so fast that no traditional book can hope to be up to date” (Kurzweil, 2024, p. 53). Some 28 months after ChatGPT first appeared, Hsu Hua, a New Yorker staff writer and Bard College professor, asked “What comes after A.I. has destroyed college writing?” (Hua, 2025, p. 21). José Antonio Bowen and C. Edward Watson have an answer in their book Teaching with AI: A Practical Guide to a New Era of Human Learning. Their answer to AI is not to ban its use but to incorporate it in such a way that the values of liberal arts education are not only sustained, but augmented. They outline examples of AI use in settings that still “look like college” with classrooms, professors, class discussions, assignments ... and grades. They focus on writing instruction, but their ideas readily transfer to broader contexts and different AI variants. Direct experience with an AI model is essential for fully grasping the possibility of a Bowen and Watson classroom. The authors use the OpenAI model ChatGPT4 as their default AI application. “GPT 4 is a substantially smarter and better AI; do not assume you understand anything about AI until you have used it for several hours” (p. 23). This test drive should include questions—now called “prompts”—to the application that require more than simple retrieval of data from an encyclopedia site. Machine intelligence has moved far beyond philosopher Daniel Dennett’s appraisal in 1994: “… most AI agents...are designed on the model of the walking encyclopedia as if all the information in the inner environment were in the form of facts told at one time or another to the system” (p. 172). With the introduction of large language models (LLMs), AI does listen and respond to natural language input (p. 15). With some sense of how AI works, the reader can then explore suggestions in this practical guide that range from basic writing instruction to advanced levels of composition. An example that would be practical for business-oriented students would be the prompt and follow-up suggested on page 63: “Using examples from the Trader Joe’s Fearless Flyer, create 25 new TJ products and write descriptions.” Students can experiment with style and voice by asking the AI to rewrite a sample of their work as a “New York Times columnist” (p. 55), “Formal academic” (p. 57), or “King James Bible” author (p. 45). Word processing now offers the same options Review of Teaching with AI 3 that “morphing” brought to graphics. Moving beyond text-only work, students might be intrigued by their AI’s offering to turn their text into a graphic novel similar to the example on orchestra conducting (p. 221). Most journalistic outcry has focused on AI’s ability to instantly generate a passable college-level essay in response to a few basic input “prompts.” This emergence of easy access to AI generated a wave of research on students who are handing in AI-produced work as their own, and a secondary wave of commercial AI- detection applications for colleges to detect such fraud. Bowen and Watson include a useful documentation of this phase of AI history and make a convincing case for “decriminalizing” artificial intelligence. Following the early chapters on the history and functionality being developed in computer science, they emphasize that AI can do much more than crank out five-paragraph college essays. The number of specialized educational AI tools (think apps on your cellphone) is about to explode. Most will be leveraging ChatGPT and other AIs in the background. Early examples are efficacious and useful. (p. 103) The new AIs, as they show with a range of examples, can create rubrics, read and grade essays, and rewrite essays in different styles. Generative AIs can turn essays into podcasts or turn audio recordings into organized essays with language translations if needed. AIs can record a professor’s lecture or read a publication by the professor and then write an essay in a style mirroring that of the professor (p. 120). The professor may have used AI to prepare the lecture or even a series of lectures formatted for publication in a journal or as a book. These and other administrative uses of AI have been publicized by reporters who manage to make their deadlines with a little help from their AI assistants. Bowen and Martin are aware that the AIs generate alarm as well as content: Starting from the recognition that AI is already ubiquitous (if not as visible as ChatGPT), student and faculty AI-literacy needs to start with a deep understanding of the potential problems, privacy, misinformation, the potential for bias, plagiarism, ethics, and exploitation. (p. 151) The authors build a plausible case for incorporating, rather than suppressing, these features. Rather than focus on rules, honor systems, weak analogies—like “Would you let someone do your calisthenics for you?”—the authors describe a method for working with students to develop creative approaches to communication. Throughout their book, Bowen and Watson reveal a deep empathy with their students. They believe in the tradecraft of teaching in higher education, but they also think educational practice must change in the era of AI. In a section on creating student assignments, they describe an initial classroom activity that uses sticky notes so that student groups can write and compare important values they hope to see practiced during course discussions. “The next step is to ask each group to create behaviors that align with their goals: what specifically does respect look and sound like in a classroom discussion?” (p. 134). No AI, in fact, no computers, are used in this initial lesson plan. Classroom discussion is the foundation of their instructional method. This is not a “distance-learning” proposition. Education Review/Reseñas Educativas/Resenhas Educativas 4 With guidelines established for class discussion, the students would study an established theory of higher-order thinking—the Bowen and Watson’s example is a revision of Bloom’s Taxonomy of Educational Objectives (Anderson et al., 2000). One of the outcomes of their course would be that students develop a deep understanding of how their work will be evaluated. Chapter 8 includes a description of how students could use AI to generate and experiment with alternative varieties of rubrics. This approach departs from a traditional course syllabus that dictates a priori rules by which grades will be calculated. One can imagine students in a Bowen and Watson classroom reassembling in their sticky note groups, this time with laptops to create different rubrics, prompting their AIs to write short papers and then examining how the different group-created AI rubrics would grade the papers. The classroom would become a lab for experimenting with writing. Students would be learning about hierarchical levels of thinking using AI tools. You can also ask AI to help design your rubric. PROMPT: Create a rubric for a first-year writing class at a community college in Arkansas. Your rubric should be in table form with the first column being the list of criteria and the first row being a sequence of points (0, 40, 60, 80 and 100). Write a one-sentence description of the quality of each criterion that will be rewarded with those points. Also note the predicted level of quality that an AI can reach for each criterion. Customizing with additional context and iterating will improve your rubrics. • Using my syllabus and course learning outcomes, design a rubric to evaluate student work in this course. • Make sure to include the following criteria. • Update the language in this rubric to reflect advances in technology. • Suggest refinements to this rubric based on best practices in my field. • Given student performance on the first assignment, how might I modify and clarify this rubric? (p. 154) During the semester they also would be learning more about the strengths and weaknesses of AI while also gaining a notion of why some writing is considered “good” and some “bad.” Or would they? Traditional writing instructors might certainly have their doubts, especially if they know a program like ChatGPT only from reading news articles that focus on plagiarism detection. Skeptical teachers handed a copy of Teaching With AI might begin with Chapter 6 “Cheating,” and then move on to “Policies,” skipping the useful introductory sections that explain the machines propelling this “New Era.” Although colleges are transitioning from banning the use of AI to requiring AI—with restrictive guidelines that include surveillance software—much academic skepticism remains. Some writing teachers believe they don’t need software to spot the AI signature. They believe generative AI has certain characteristics—overuse of the long “em dash” for parenthetical phrases and excessive use of words such as “plethora” that students do not ordinarily use or understand—usages that betray the presence of AI assistance. Inevitably, false positive test results occur, whether from human or machine examination. Bowen and Watson’s attitude towards plagiarism Review of Teaching with AI 5 detection comes across in their introduction to the section on cheating—a parody of the old radio-TV program Dragnet (p. 106). Accompanying the Bloom cognitive hierarchy, Bowen and Watson invoke another familiar learning theory—that of Lev Vygotsky, who proposed that learning develops with tutorial assistance through “proximal zones” of skill and problem solving (p. 77). In place of a student-tutor relationship, the Bowen and Watson AI model classroom pairs a student with a personal AI writing coach—perhaps ChatGPT in a customized edition. Although ChatGPT at this point produces what the authors call “average” or C-level work, they feel that the ability of AI to quickly generate many alternative perspectives will have the effect of advancing a student’s levels of thought. They cite a frequently mentioned sophomore at Harvard who took real assignments from her eight classes and then asked her professors to (re)grade them, telling them they might be hers or might be ChatGPT. They were all written by ChatGPT and received a 3.34 GPA—note: that’s an average grade at Harvard (Bodnick, 2023). With a little editing (possibly even from AI-generated suggestions) a student could produce a B paper with minimal effort. It’s again that Vygotskian notion of the Zone of Proximal Development (see chapter 4); students are receiving help from a knowledgeable, though AI, peer. The question as you grade may be: In what ways has the student moved above and beyond what AI produced for them. (p. 151) A skeptical reader may question the practicality of certain suggestions in Bowen and Watson’s guidebook. One of the most-noted problems with AI is its tendency to created false information, usually called “hallucinations.” The authors’ answer is to turn this glitch into a feature. Hallucinations (both AI and human) may be dangerous, but they are also a feature of creativity. Originality is about thinking the unthinkable, and appear on track to do this better than humans. It is hardly a surprise that many artists also display occasionally odd and antisocial behavior: it might be a feature of creative thinking. (p. 68) In inter-university communication, “creativity” might be held in check—facts in assertions need to be confirmed, not elaborated! An excerpt from the section on administrative uses for teachers includes this suggestion: You could, for example, get an AI to draft an accreditation report, optimize your class schedule, act as an external consultant for your strategic plan, create a departmental dashboard, plan an event, anticipate future student demands, review government compliance, create a department newsletter, do a sentiment analysis of teaching, or review policies for equity and recommend changes that would increase graduation rates or support for underrepresented students.... Try uploading your department goals and asking how it might better align with the university’s strategic plan: maybe make this an activity for a faculty meeting. (pp. 104-105) Education Review/Reseñas Educativas/Resenhas Educativas 6 Regarding the last suggestion, a creative faculty person should anticipate the abundance—a plethora?—of ideas presented at the meeting. Suppose all persons in attendance bring their own AI-assisted lists of ideas. The natural thing for the department chair, of course, would be to ask that the ideas be submitted as a pool of suggestions to the department AI for an executive summary. A section on AI faculty assistance suggests a use for a conference with a student: Now imagine that, like the AI-assisted doctor, you have an AI that recognizes the student and immediately organizes the relevant information and even prompts you to ask how the student is enjoying the history course you recommended. The AI listens to your conversation …and prompts us that the student will need Requirement 101 to complete her major. ..., AI will also be better at recognizing whether the student needs other assistance: a zumba class, recommendation for a campus club, or counseling for stress (based upon previous conversations of interests that you have long forgotten). (p. 91) Students may have been cautioned about sharing too much personal information to any online data source. The student may also be recording the conversation to get clues as to what the instructor expects for a grade of A. An additional issue might be that some students have been using a particular AI for the past two years and established a sort of human-computer bond with the AI that they don’t want to share or use for school work. A typical alarm about AI use outside the classroom comes from stories of young people who received bad advice from their AI confidants. Perhaps students would ask their personal AI partner to evaluate the comments from their Vygotsky AI tutor and suggest whether or not to participate. Students are likely to have had two or three years of experience with AI before enrolling in the college course. As Hua (2025, p. 21) points out in his New Yorker article: ‘Any type of writing in life, I use A.I.’ he said. He relied on Claude for research, DeepSeek for reasoning and explanation, and Gemini for image generation. ChatGPT served more general needs. Regarding the problem of assigning grades, suppose a student spent some time recording her thoughts on an application such as Apple Speak, or Otto, getting back an organized summary of their ideas. The student could then use the transcription summary as prompts to ChatGPT. The result would be the student’s ideas in the formal attire of standard English. Consider prompts as recursive computer programming: a student might use ChatGPT to suggest prompts and then feed these prompts back for further development. What rubric would help determine a final grade? To reprise the campanology conceit at the beginning of this review, note that Poe’s bells chimed “in a sort of Runic rhyme.” Given the aphorism “History doesn’t repeat itself, but it rhymes” (O’Toole, 2017, p. 358), we should look for rhymes from previous eras of technological innovation. For example, in Computers as Theater, Brenda Laurel (1993) expanded the concept of the computer from the narrow idea of a productivity tool to a conceptual medium, and suggested Aristotle as a useful resource. In Why Things Bite Back: Technology and the Revenge of Unintended Consequences, Review of Teaching with AI 7 Edward Tenner (1997) pointed out how the insertion of new tools in a creative sequence distances the author from the work being created. The recurring argument that machine intelligence lacks “common sense” is now being intensively addressed in the quest for general artificial intelligence (Anicker & Flashoff, 2024). Connecting AI to tools in the human world—one interpretation of “machine agency” (Lanham, 2025)—has been a recurring theme through science fiction. Bowers and Watson mention Star Trek, but of course, there are many others (Asimov, 2013; Capek, 2001; Conklin, 1964; Gibson, 2020). Bowen and Watson have taken a practical approach to the larger issues facing the uses of machine intelligence: It is beyond the scope of this book to discuss whether AI will ever produce the great American novel or how tragic it will be for individuals to lose their jobs to AI. (p. 76) A practical guide does need to set limits to speculation, but these are still two serious topics. The great American novel—a familiar phrase attributed to John William DeForest (1868), who wrote that such a work “… has seldom been attempted, and has never been accomplished further than very partially” (p. 28). Aside from pondering the dilemma of awarding the Pulitzer to a future version of ChatGPT, there is a much greater possibility that students using AI assistance will produce creative work that is exceptionally good. A suitable student assignment would be to explore the possibility of creativity in a “post-human” world, an issue discussed in depth in Slavoj Žižek’s, Hegel in a Wired Brain (2020). The tragedy of individuals losing their jobs to AI should strike college students as highly relevant, even as they plan to find work as experts on writing AI prompts. They may see the irony of being hired to help program machines for replacing human employees. The tragedy—quite real, in fact—could extend to jobs in universities and civil service, including legal services. The previous “post-human” question applies here: could machines do a better job? Or if not better, maybe just cheaper? Bowen and Watson anticipate this question at the beginning of their book: It is the job of educators to help students become better thinkers. Our new job is to help them become even better thinkers with AI. It is essential that educators start to talk about these issues with students; if we want students to use AI responsibly, both in school and beyond, AI ethics must be baked into curriculum and include AI literacy, an emerging essential skill. (p. 3) References Anderson, L., Krathwohl, D. R., Airasian, P. W., Cruikshank, K. A., Mayer, R. E., Pintrich, P. R., Rathes, J., & Wittrock, M. C. (2000). Taxonomy for learning, teaching, and assessing: A revision of Bloom’s taxonomy of educational objectives, complete edition: Pearson. Anicker, F. Flashoff, F. G, (2024). Common-sense attributions of AI agency: Evidence from an experiment with ChatGPT. In M. W. Bauer & B. Schiele, AI and common sense: Ambitions and frictions: Routledge. https://doi.org/10.4324/9781032626192-18 https://doi.org/10.4324/9781032626192-18 Education Review/Reseñas Educativas/Resenhas Educativas 8 Asimov, I (2013). I robot. Voyager Classics. Bodnick, M. (2023, July 18). ChatGPT goes to Harvard and does better than you think! Slow Boring. https://www.slowboring.com/p/chatgpt-goes-to-harvard? Capek, K, (2001). R.U.R. (Rossum’s universal robots). Dove Press. Conklin, G (1964). Science fiction thinking machines: Bantam Books. DeForest, J. W. (1868, January 9). The great American novel. The Nation. Dennett (1994). Language and intelligence. In J. Khalfa (Ed). What is intelligence? (pp. 161-178). Cambridge University Press. Gibson, W. (2020). Agency (The Jackpot Trilogy). Berkeley. Hua, H. (2025, June 30,). The end of the essay: What happens after A.I. destroys college writing? The New Yorker, pp. 21-27. Kurzweil, R. (2024). The singularity is nearer: When we merge with AI. Viking. Lanham, M. (2025). AI agents in action. Manning. Laurel, B. (1993). Computers as theater: Addison-Wesley. OpenAI ChatGPT 4o. (2025, May). Heidegger Sein Dasein Shift. Personal communication, May 2025. O’Toole, G. (2017). Hemingway didn’t say that: The truth behind familiar quotations. Little a. Poe, E. A. (1992). The bells. In The collected tales and poems of Edgar Allan Poe (pp. 954- 955.) Modern Library. Tenner, E. (1997). Why things bite back: Technology and the revenge of unintended consequences. Knopf Doubleday. Žižek, S. (2020). Hegel in a wired brain. Bloomsbury Academic. About the Reviewer David Fowler is an emeritus professor of mathematics education at the University of Nebraska–Lincoln (UN–L). He has an undergraduate degree in mathematics from Harvard College (1962) and PhD from UN–L in mathematics education and instructional technology (1991). His primary interests are in finding uses for emerging technologies in mathematics education, especially with the application Mathematica, developed by Wolfram Research, Inc. He has been a Wolfram Visiting Scholar at the WRI campus and was Editor-in- Chief of the journal Mathematica in Education and Research (Springer Verlag) from 1998 to 2002. At UN–L he chaired 19 doctoral students and 74 master’s students in mathematics or technology education. In 2011, he received the Milton Beckman Lifetime Achievement Award for Advancement of Mathematics Education from the Nebraska Association of Teachers of Mathematics. He first encountered the term “artificial intelligence” in 1960 in a course on mathematics and linguistics taught at Harvard College by Anthony Oettinger. The central problem posed in the course was machine translation, specifically from Russian to English. A promising approach at that time was to apply Markov processes to the new theories of syntactic structure being developed by Noam Chomsky. https://www.slowboring.com/p/chatgpt-goes-to-harvard? Review of Teaching with AI 9 Education Review/Reseñas Educativas/Resenhas Educativas is supported by the Scholarly Communications Group at the Mary Lou Fulton College for Teaching and Learning Innovation, Arizona State University. Copyright is retained by the first or sole author, who grants right of first publication to the Education Review. Readers are free to copy, display, distribute, and adapt this article, as long as the work is attributed to the author(s) and Education Review, the changes are identified, and the same license applies to the derivative work. More details of this Creative Commons license are available at https://creativecommons.org/licenses/by-sa/4.0/. Disclaimer: The views or opinions presented in book reviews are solely those of the author(s) and do not necessarily represent those of Education Review. https://creativecommons.org/licenses/by-sa/4.0