Teaching with AI: A Practical Guide to a New Era of Human Learning by José Antonio BowenMy rating: 4 of 5 stars
If you are an educator, no matter where you land on the topic of AI, you should, as Louis Epstein said in a review of the first edition in the Journal of Music History Pedagogy*, “read this book.” Epstein also notes the authors’ “almost gleeful boosterism,” and indeed, there is no real doubt that they assume AI as a given. In the epilogue to the second edition, the authors generously allow: “Faculty are the ultimate decision-maker, arbiter of quality, and content expert within their own classroom” (319). But make no mistake, the book is unapologetically prescriptive: “You need to start playing and working with AI” (319); “Our curriculum and classroom discussions also need immediately to include AI topics around ethics, data privacy, civics, environmental and energy costs, labor, discrimination and bias” (321). On those latter fronts, if you are looking for advice on how to approach those topics, this is not the book for that (it will only tell you that you need to, not necessarily HOW to). Most of the book is focused on better prompt writing, because, as the authors claim: “All assignments are now AI assignments” (245). In his aforementioned review, Epstein notes that the “unspoken thesis” of the book is that student and faculty are now forced “to decide what really matters to us in our teaching and learning”. I don’t disagree, actually, but I have a feeling my human-generated SWOT analysis of AI and that of the authors would differ greatly.
Bowen and Watson offer: “The new goal of any assignment, especially a writing assignment, is to emphasize the important human contribution while recognizing the realities presented by AI” (250). I like the use of the word “realities” here, but the book is instead heavily focused on “opportunities”, with a strong belief that if we invest the time, students can integrate AI in meaningful ways. What isn’t clear (yet), is how this will really play out, given students’ various motivations for “efficiency.” There are good select studies on issues such as human agency in creative problem solving in AI collaboration, but the full-scale adoption of AI as our “thought partner” in higher ed has not yet come to pass in such a way that we have large-scale data on the impact. But we are probably getting there.
There are some seeming contradictions. Some of the examples and prompts offered by the authors mention how AI can be used to alter/experiment with “voice” in a piece of writing. I’ll be transparent and say that this, along with a new interpretation on the role of citation in writing, is one my biggest issues with AI. If I go full-on “the Borg are coming,” this is where I see a real abdication of human-ness. So, I found it odd to offer guidance on including “voice” in prompts for AI (p.59), but then in Chapter 8, on “Cheating and Detection”, the authors advocate starting a conversation with a student who you suspect might have used AI with “I don’t really hear your voice as much as I would like” (161). I happen to think that’s a GREAT way to start the conversation (and I have), but I also don’t encourage them to use AI with thorough prompts that include directions to “respond as if you were x” or write “like an engineer.” The conflation of “language” and “voice” I find problematic as well.
There is also a lack of acknowledgement of potential tokenism and essentialism in prompting AI in the ways the authors describe. For example, as a faculty member, the authors promote doing a practice conversation with AI wherein it is prompted to be “my student Jeff, who is a nineteen-year-old from Wisconsin majoring in biology and taking my course pass/fail” and then directed: “Please respond as if you were Jeff.” (133) While only a “practice conversation” and not a script, it posits poor Jeff as some sort of archetype of whatever identities we’ve given the AI. It isn’t to say that our experience of Jeff IRL is complete, but one hopes that we might have a better sense of Jeff as a student than an AI combing known character traits of nineteen-year-olds, biology majors, and folks from Wisconsin. More problematic (perhaps) is the counsel: “Perhaps you simply ask the AI to be an under-represented voice in the room (225).” On the same page, an author might ask an AI to “reimagine” their work with the lead character as “an Asian American and identify what plot lines might need to be changed.” I note that it does say “might”, but the idea that we would ask AI to represent “an Asian American” doesn’t seem any better than asking an Asian American to speak for all Asian Americans.
My critique of the authors’ suggestions for citation goes far beyond the confines of their book. There seems to have been a large-scale shift in discussion about “checking citations” without highlighting the point of citation in the first place. I carefully say “seems” here, because this is a baseline perception, not a carefully researched thesis. But the authors tell me that I might, as a scholar writing an article, prompt an AI with “Who are the other major figures in this field, who might be reviewers of this article? What work of theirs should I cite?” (114). I understand the pragmatism there, but why not instead: “What work of theirs should I read?” I don’t think that’s a pedantic distinction.
In the chapter on “Reimagining Creativity”, the authors supply a cartoonish image created with Gemini-2.5 via a prompt that prescribed “style of a Renaissance woodcut with seventeenth-century London in the background” and features Genghis Khan on the left with his face merged into a strawberry, and Queen Victoria is on the right with her face hybridized with a cauliflower (Figure 4.1, 81). Both figures then have a thought bubble (as directed in the prompt) that asks “Will anyone care if we are AI generated?” The image is an answer to the question the authors pose: “What if AI makes the art itself?...Humans still need to think of the ideas, but does it matter whether AI was used to realize the actual image?” (80) Bowen and Watson don’t have the answer, but only offer “There are important personal, social, economic, and ethical issues that deserve a place for discussion in society and in our revised curricula.” (80). These discussions have long been part of artistic and creative production (think digital photography), so they aren’t wrong, but I’m growing tired of the “just because we can” advocacy when it comes to AI. I'm reminded of an AI demo on my campus wherein we were shown how one might create a "nature documentary" using AI generated images. I was alarmed that the irony was not acknowledged.
Ultimately, however, the book backs a lot of what is just good pedagogy. For example, there’s an excellent suggested exercise for a class on defining values, teamwork and accountability (180 – 181), something that mirrors the “Community Agreements” I facilitate for all my classes. Whether that discussion needs to be in service of using AI or not might be challenged, but it is good that it is offered in the book. Table 12.1, an “Assignment template combining motivation, task clarity, and success criteria” is another useful example both in and outside of AI contexts. And while I have not yet embraced intentional integration of AI into my own teaching/course content, I might get behind allowing specific prompts such as “Identify which ideas and arguments in this essay are common, flawed, repetitive, heteronormative, or culturally limited”(268). I see this as a bit different than the prompt I mentioned above that essentializes Asian Americans, but I’d have to run it on several examples for myself to see what a given AI calls “culturally limited.”
And that’s probably the biggest problem. One takeaway from the book is that converting core critical thinking skills into AI prompts is no simple task, and it isn’t until the Epilogue that Bowen and Watson seem to vaguely acknowledge the immense amounts of labor being placed on faculty: “Much of the AI-fueled pedagogical revision on college campuses has emerged as another unfunded mandate for faculty and their time.” (325) Even if I entertain sharing just the basic information of this book regarding how to prompt, I cannot imagine doing so at the sacrifice of things I find just as (or more) important for my particular course. And I certainly do not have the time to teach them “how to be expert prompters” AND music history. It isn’t to say that I can’t incorporate “relevant” examples and creative projects, but I certainly cannot afford the time to invest in the type of nuanced prompting the authors promote. So, who will? That remains to be seen.
So yes, my colleague Louis is right. Read this book. It features excellent and recent sources, and it will either strengthen your resolve, confirm your intentions, or move the needle to the middle, depending on your starting point.
No comments:
Post a Comment