We Shouldn't Destroy Ourselves Fighting About AI

Bart Fish & Power Tools of AI / https://betterimagesofai.org / https://creativecommons.org/licenses/by/4.0/

How we talk to one another about AI needs to improve. Last October, my colleague Bob Cummings went to India to attend the Mind & Life Institute’s 39th Dialogue on minds, artificial intelligence, and ethics. Emily Bender was also present, along with many other voices in the AI space. As the conference came to a close, the attendees met the Dalai Lama! When I asked Bob what he learned from the conference after meeting the spiritual leader of Buddhism, he said, “We shouldn’t destroy ourselves fighting about AI.”

The current AI discourse across social media is pretty dark. Well-meaning people are angry about AI in schools and broader culture. It’s understandable given the changes AI has wrought. But our emotional reactions about AI aren’t necessarily leading to productive strategies to deal with it. I’m reading the Pope’s encyclical on AI and hearing from local ministers about what machine intelligence means to them. In those conversations I’m struck by the level of compassion and focus on human agency and dignity that’s often absent in the heated AI debates across social media.

One of the many reasons I chose to coauthor The Norton Guide to AI-Aware Teaching with Annette Vee and Derek Bruff is because I believe that AI has caused us to lose touch with teaching as a caring profession, one built on relationships and shared values. What I’m seeing increasingly in the public discourse about AI in education is folks dividing into camps and talking past one another, when now is the time for us to work together.

There’s no better place than a college campus to show how people with many diverse and often contrary opinions can come together to find common ground. We need to work now on how we talk to one another about AI, focusing on seeking compassion in these increasingly heated discussions about the very real challenges posed by this technology. Nearly every campus must develop a common language, an agreed-upon set of principles, a social contract, if you will, about how we contend with AI.

What we need now is a shared campus commitment to helping students and faculty understand and examine the capabilities, limitations, ethics, environmental costs, labor implications, and disciplinary uses of generative AI. We need Critical AI across the curriculum. This doesn’t mean every faculty member must use AI. Far from it. Rather it guides campuses in developing a common language for deciding when AI belongs, when it does not, and what values inform those choices. As Josh Brake argued “we need not integrate it. It’s a subtle point, but engaging with AI allows us to experiment with AI without giving in to a rhetoric of inevitability.”

Critical AI across the curriculum helps us move from fractured and individualized responses to a collective set of principles to serve as a guide to navigating AI.

Enabling AI Access Isn’t Condoning AI Use

Many people are decrying the University of Chicago’s decision to activate an enterprise version of Anthropic’s Claude across their campus. Some are saying it sends the message to students that essentially green-lights the use of AI in the classroom. However, if you read the President’s reasoning behind the decisions, you find something far more nuanced and human:

Policies vary by course, and you are required to follow them closely, but the sum across your full curriculum will give you diverse learning experiences with and without AI assistance. Instructors at all levels are navigating a fast-evolving landscape, and the University has a duty of care to ensure that the education offered to you is responsive to these technological developments by teaching you how to think with machines, how to think without them, and how to think about them.

That sounds like the correct path to me. Each campus should consider how they engage thoughtfully and intentionally in conversations and decisions about AI, warts and all. Administration should create the conditions to support faculty in this effort and move from initiatives solely focused on AI usage to creating environments where faculty and students alike practice discernment when it comes to the technology, making reasoned arguments about when and whether AI should be used.

I’ve written before about the equity and access issues posed by our students using free AI tools, but let me put some sobering statistics to those numbers. The Mississippi AI Network (MAIN) recently shared the number of users of ChatGPT at each public university and college throughout Mississippi. At my institution alone, over 16,000 students have signed up for ChatGPT. Of that number, more than 1800 are paying for some level of greater access beyond the free version. That means students on the free plan are sending their data, along with their professor’s assignments, slides, and other teaching materials into the commercialized version of the chatbot, and are also paying OpenAI hundreds of thousands of dollars each year for premium access to better models not offered on free plans. Enterprise plans protect data and allow students and faculty equal access to premium AI tools. Offering such access does not mean administration is telling students that their usage of AI is appropriate. It gives everyone the opportunity who wants to explore AI the assurance that their data is safe and access is equal for all.

Our Personal Values Inform And Create Our Campus Values Around AI

A question many of us have been asking is what our campus’s approach to AI should be. Great question! An even more pointed one would be, what efforts have we made to come together as a community to tackle such questions? The majority of higher education’s response to AI has been individualized and focused on course-level policy. What we really need to do is reframe discussions from AI usage by students to shared values and guiding principles.

Currently, the norm across most institutions is to avoid any sort of uniform stance toward AI. Campus leadership has long allowed and encouraged faculty to set their own policies and preferences, allowing them to have the freedom needed to select pathways and approaches to teaching content and choosing what technologies are appropriate for their students. Academic freedom matters here tremendously. What makes AI so difficult is its ubiquity and its ability to be applied across disciplines and roles. Faculty are being asked to take individual stances on a general technology they struggle to comprehend and make pedagogical decisions based upon it without direction from their institutions, because few have taken the time to create the spaces necessary for the types of conversations necessary to develop a common understanding.

This has led to fractured responses and bewildering experiences for students. Come fall, a first-year university student will likely experience certain classes that ask them to leave their phones and laptops and use pen and paper before shifting to classes mere moments later in large lecture halls where everyone is constantly on a device. All in the same day, on repeat, not just for a semester, but for their entire college experience. No one believes this is ideal, and the overwhelming response I’ve heard from people is that higher education’s fractured response is just temporary. Temporary until what?

The Good Work Getting Lost in the Shouting Matches Over Machines

We’re getting so lost in the digital discourse that many of us are forgetting that AI doesn’t simply need to happen to us! We can shape how we use or refuse technologies, but it takes highlighting the good work individuals and campuses are doing to make certain students and faculty can navigate AI and not getting dragged down in pointless fights about it.

Teaching Students About AI

At Harvard, Jane Rosenzweig and Tad Davies created a required module for all students entering writing courses that asks them to critically engage with “how large language models work, their impact on education, and broader questions about AI and society.” The notion that students should first learn about AI before any discussion of using AI is so foundational and vital in creating a culture where students practice discernment around these tools that institutions across the country should work rapidly to establish similar programming.

Being Intentional When Talking About AI

If you read this newsletter often, you probably notice that I prefer to use images people have created vs. generating them. That wasn’t always the case, but when I came across the Better Images of AI Project through Katie (Kathryn) Conrad ’s critical AI work, it struck me that a pretty basic and intentional action I could start taking on my part was selecting images that move beyond cliches and sci-fi tropes.

Helping Students Construct Values in Response to AI

In the K12 space, the Humanities Department at the North Carolina School of Science and Mathematics established the following learning outcome for their students:

Build critical AI literacy by examining the basic functionality, limitations, and applications of LLMs in Humanities-based learning tasks; evaluating the ethical implications of AI; and developing their own personal values around AI usage in and outside of the classroom.

That idea of values is so central to this question of using AI or refusing it. What better way to ask students to construct their own set of values around AI, instead of just giving them a policy embracing or forbidding it?

Create a Common Language Around AI Usage

At the Ohio State University, their Committee on Academic Misconduct (COAM) developed a system of standardized “Academic Integrity Icons to promote transparent communication with students about the use of various resources on assignments, including AI.”

Imagine that! Logging into your LMS and being able to easily communicate with your students what is acceptable for an assignment using a common set of icons they are all aware of! We should all investigate creating a common template that everyone can use to ensure faculty AI preferences are communicated transparently to students.

Create Space for Real Conversations

Even simple gestures, like sharing a drink and talking about AI’s role on campus, can help faculty develop common ground over what boundaries they’d like to set. Guy McHendry held an “AI Speakeasy” for faculty and found broad agreement about not wanting to use AI to grade student work.

Take Stock of What Has Changed, Then Move Forward

Before we start considering the opportunities AI may hold, we need to first take stock of what has been lost, or at the very least, made vastly more challenging. In writing, students no longer have to face the fear of the blank page and the beneficial friction involved in the process of writing. A machine can whip up a first draft for you, but the question we’re faced with is whether that’s really what students need, and if not, what can we do to preserve the rough draft?

Beyond writing, all the major foundation models can solve most math and biology problems, and more concerningly, generate the process faculty look for as evidence in learning. Some faculty have turned to changing the environment to make learning more visible by going to studios, labs, or other embodied learning. This may be AI-resistant to a point, though; all it takes is the microphone or camera on a smartphone to turn an analog assignment into a digital one that AI can finish.

It’s increasingly looking like an AI bubble isn’t going to happen, at least not in the way we think it should, so now what?

Few campuses can can afford to upend all instructional practices in an effort to AI-proof learning. The movement to secure assessment against AI is well-intentioned and needed, but glosses over the extraordinary costs associated by shifting assessment to in-person, or training faculty to proctor them, or establishing the needed resources to meet the rising number of students in need of accommodations as a result. Students need opportunities to fail and learn from those failures. Shifting to strictly secure assessment would require faculty to offer students multiple opportunities to retake an exam and that’s in addition to considerations if such secure assessment types authentically capture what students need to know. Institutions will have to increase tutoring and student support services, including mental health support because such high pressure testing culture will be sure to elevate student anxiety. We cannot move to an all or nothing, AI or no AI system.

Those faculty with the time to change their assessments generally teach a reasonable teaching load and have security. Online faculty, those who teach a heavy, multi-section load of courses, or contingent faculty, cannot reasonably respond to AI because of the modality of their position and conditions of their labor. Those faculty who have altered exams often assess less evidence of student learning than before as a result of the increased workload. There is also only so much you can assess in a timed bluebook, oral exam, or traditional test. I suspect the recent news of seeing an uptick in students receiving more As may, in part, come from faculty becoming less rigorous on students when they can be assured their learning is human and not the result of a chatbot.

But the challenge is multi-layered and not confined to students or learning outcomes. Educators at nearly every level are having to ask themselves, just as students, if AI can complete a task for me, then why shouldn’t I use it? Faculty usage of AI should be intentional and transparent, yet how many of our campuses have such policies in place or procedures for ensuring faculty follow them? Have we even started this conversation?

A teaching assistant starting out today no longer faces the terror of a blank syllabus or an empty assignment bank, as AI is present in nearly all course management software. A competently generated slide deck with accompanying activities built for a 50-minute or 75-minute class section is an easy task for any foundation model to produce in mere seconds. Most instructors take years to develop the instincts and skills necessary to develop sound teaching materials. The question becomes what role machine intelligence should play when one is learning how to structure and teach a college course for the first time, or for the 30th time.

It takes time to learn how to teach well. Anyone who has created assignments and activities for a class knows that failure is part of the process. It is humbling, frightening, but it lets us know how to become better teachers. What critical feedback will faculty receive from students when their assignments are designed by a machine trained to please?

A newly hired adjunct a few days before the start of a term will likewise need to navigate what role AI may play and consider the time needed to create a thoughtful course organically or generate one. A great many faculty in contingent roles have refused to use AI, but I’m certain others have asked what the point is in spending the time and energy developing materials for institutions that consider their labor to be, by design, easily replaceable.

Amid all of these challenges, there is the possibility for good uses of AI to emerge, and for use to move beyond just critical conversations about AI and start seeing creative possibilities. That’s only going to happen through the sustained commitment of campuses that center human beings and all of our messiness over efficiency. A new teacher or overworked adjunct could find valuable help in using AI for a great many tasks. Students may likewise develop the type of thoughtful, critical, and intentional relationship with AI that moves past uncritical adoption. Yet, none of that will arise unless we support the tremendous work taking place to get past the noise. Nearly all of us are experiencing valid emotional reactions to AI. Some are excited, many are shocked, many more are grieving, and I imagine the great majority are feeling some mix of all at once. Find productive ways to talk to one another about it!

Preorder The Norton Guide to AI-Aware Teaching

Thanks to the wonderful team at Norton, the guide is now available to pre-order! My coauthor, Derek Bruff, wrote the following in his newsletter. The ebook is expected to be available on July 1st, and print copies are expected to start shipping on September 24th. Here’s how you can get a copy:

1. Our publisher Norton is pleased to offer the guide as a free ebook for all instructors currently using a Norton textbook. If that’s you, you’ll receive access from the Norton team when the ebook is available July 1st and can contact your local Norton representative with any questions.

2. If you would like to pre-order the ebook so that you have it July 1st, you can now do so through Amazon and Barnes & Noble and perhaps other retailers.

3. If you would like to pre-order the paperback version of the book, you can now do so through Norton, Amazon, Barnes & Noble, and likely other retailers. If you go through Norton, be sure to use the code AIFREESHIP at check out to get free shipping!

4. If you would like to order multiple copies for a campus reading group or some other faculty development effort, Norton has an option for you: On orders of 10 or more print copies, we offer 50% off the list price and free domestic shipping. (Such orders must be on a nonreturnable basis.) To take advantage of this offer, contact Peter Wentz at pwentz@wwnorton.com with subject line “Norton Guide to AI-Aware Teaching.”