Publication Announcement: Building AI Literacy: Critical Approaches and Pedagogical Applications

Volume 48, Issue 2: Generative AI’s Impact on Education: Writing, Collaboration, & Critical Assessment

Yutong Liu & Digit / https://betterimagesofai.org / https://creativecommons.org/licenses/by/4.0/

For the past year and half, I’ve been working with my colleague Stephen Monroe on editing a special issue for Thresholds in Education about generative AI’s impact on education. The second of three issues for this special volume is now available and may be accessed freely. Copied below is our introduction. We finished drafting it in May and then the summer and fall semester caught up with us. We didn’t have time to return to it. I’ve added some areas in bold to try and update it, but as a sign of the times we live in, much of what we wrote feels like we’re simply tying to address the moment we are living in.

Once again, we are so grateful to have the opportunity to elevate voices who are confronting GenAI in our classrooms!

Issue 2 Introduction: Building AI Literacy: Critical Approaches and Pedagogical Applications

Surfing in a Tsunami

This is a time of intense technological change: an AI wave. An AI tsunami? Eric Schmidt’s advice—to teachers and to everyone else—is to lean in, learn, and adopt AI as quickly as possible. He warns that we ignore AI at our peril. It sounds like Silicon Valley hype, but Schmidt is probably right. OpenAI’s ChatGPT now has over 400 million weekly worldwide users. That’s impressive growth for a tool that launched little more than two years ago. And, of course, OpenAI is only one company causing wakes. Anthropic, Microsoft, Meta, X, and Schmidt’s own Google are also thrashing ahead, delivering new models, recruiting millions of users, and promulgating the many potential benefits of Generative AI. The resulting waves—in education and in society—are unprecedented and gaining speed.

But tsunamis (and even smaller waves) can be destructive. As educators, we are right to ask skeptical questions and to worry about potential risks. For example, how do these tools threaten student learning? Where is the line between AI assistance and AI cooption? What are we losing in our classrooms, if we adopt, adopt, adopt? Some worrisome hints can be found in studies conducted by the AI companies themselves.

In 2025, Microsoft examined GenAI usage in the workforce by surveying over 300 knowledge workers about how they used Microsoft Copilot. The authors found an alarming connection between heavy usage of GenAI and a negative impact on critical thinking skills. The authors offer a thoughtful warning: “While GenAI can improve worker efficiency, it can inhibit critical engagement with work and can potentially lead to long-term overreliance on the tool and diminished skill for independent problem-solving”. Their suggestion was not to ban the tool or deny workers access, but to ensure GenAI interfaces are designed to support critical thinking—not to offload it altogether.

Anthropic’s Education Report examined one million anonymized student chats to see how university students use their flagship AI model, Claude. The results are quite surprising. Some of the heaviest users were in STEM fields and nearly half of the top chats were students using GenAI to seek direct answers or shortcuts, instead of using the model as a tutor. Dominant were queries like “solve and explain statistics problems,” “answer earth science questions,” and “solve chemistry problems.” The authors of the study provided a nuanced understanding of student usage: while some students used Claude as a tutor and collaborative partner, many more sought to quickly offload their intellectual labor. The authors warned, “there are legitimate worries that AI systems may provide a crutch for students, stifling the development of foundational skills needed to support higher-order thinking”. To that end, Anthropic’s new offering, Claude for Education, introduces “learning mode,” which uses GenAI to guide discovery and Socratic inquiry rather than simply providing students with a speedy answer. This is good. We should recognize when a GenAI developer responds to troubling trends in their own research and makes an effort to address those concerns through redesign. Such efforts make our job as educators easier and prove that AI companies and teachers can work together on behalf of our common “customers.” Such tailored AI products show that is possible to encourage students to pursue ethical and transparent GenAI usage, while also promoting learning. (We wrote this in May. Now every major foundation model has incorporated some form of a ‘learning mode’ and many don’t seem to be living up to the promise, or hype.)

But not all developers are spending such time on human learning problems or creating throttled interfaces, which may or may not prove popular/profitable. Thus far, OpenAI’s response to concerns about human learning has been muted. In lieu of redesigning their interface for students in their ChatGPT Edu offering, they have, instead, chosen a promotional path: giving away two free months of ChatGPT Plus to any college student in the US or Canada, with the slogan “ChatGPT is here to help you through finals” . Within weeks of the announcement of OpenAI’s free trial, xAI announced two free months of Super Grok for college students, and Google made the stunning announcement of offering over a year’s worth of free access to the premium version of Gemini, Google’s most powerful AI model. Market share matters very much to these companies.

Indeed, GenAI companies realized early that the power users of their products were students, not working professionals or any other demographic group, and they are now targeting students around the world quite aggressively. Their hope, of course, is to win users and establish loyalty. Popularity and profits are taking precedence over what is best for student learning. Mainstream AI companies are not directly marketing the concept of academic dishonesty, as are some marginal providers whose ads are likely prevalent on your students’ social media feeds, but mainstream AI companies are facilitating such behaviors, which are obviously adverse to traditional learning. Where are teachers and professors in these conversations? How do we encourage GenAI companies to better align their marketing with their lofty public pronouncements about societal benefits?

Those are questions sloshing around in the tsunami. In a related concern, the competing announcements of free trials of premium GenAI models only for college students may also be a blow to equity and access. If GenAI is to become a universal tool to uplift humanity, as promised by the most eager optimists, then why grant free access only to college students? Giving the most powerful versions of AI only to people around the globe who have gained access to higher education is hardly an equitable framework for scaling this technology throughout society. A shrewd and calculated marketing move to hook users at a young age? Yes. A pathway toward empowering all people equally? No.

Imposed strategies related to rollout and marketing on the part of OpenAI, Google, and xAI will surely cause continued disruptions on our campuses. These are the ingredients of the tsunami, and the bots are too often separating human teachers from human learners. Faculty are left to forge temporary order from chaos and change. For example, without stability or standardization, we are now engaging many different learners: those who don’t use GenAI, those who use the free versions of the tools, and those who have adopted trials or paid for access to premium features. The students are all sitting together in our classrooms. They have their laptops open, but they are difficult to tell apart and, of course, some of them are hiding their status.

The answer now and into the foreseeable future is open communication between teachers and students. If the bots are wedging themselves between us, we must name and nudge the bots. We must appeal to our students, rather than police them. It is an ethical path forward—and perhaps the only practical one. Faculty that rushed to adopt AI detection found the tools to be unreliable, and some innocent students have been caught up in this part of the tsunami. Perhaps detection will become possible in the future. In truth, the mainstream AI companies could enable watermarking tomorrow and give teachers this useful tool for transparency and tracking. Google has open-sourced watermarking through cryptography via SynthID, but few other AI developers have embraced such methods, which would contradict marketing efforts aimed at children and young adults. Even so, the promise of other detection schemes persist—stylometry, linguistic fingerprinting, and most recently process tracking in the form of Grammarly Authorship and Turnitin Clarity. Experts in academic integrity remain unconvinced about process tracking and its ability to curb AI misuse, noting recent advancements in GenAI Agents and Deep Research tools “have now made this strategy unreliable”. Many of our colleagues want reliable policing tools, and some of our colleagues believe falsely that they can “spot” AI on their own, but, in truth, there is no reliable AI detection available right now.

We do not see the point in undertaking a detection arms race when it comes to GenAI. Rather, we plan to spend our energy guiding students by talking to them directly about their GenAI usage in practical terms. Viewing Generative AI as solely a cheating tool cripples any potential conversation about AI’s impact on student learning and, more broadly, on their future lives as professionals and as human beings. We want to empower our students with sustainable lessons. We do not want to discourage our students with temporary punishments. Now is the time for talking with students about how this emerging technology will affect them and our shared society. So many interesting, novel, and motivating questions are suddenly available! The conversations will not be possible unless faculty can trust students, and vice versa. As bell hooks observed, “the classroom, with all its limitations, remains a location of possibility” (Teaching to Transgress, 1994). Perhaps AI companies are setting new boundaries and creating new problems, but we, as teachers, can and must protect the all-important human relationships within our classrooms.

What strikes us with alarm is by not talking with students about their GenAI usage, academia is ceding that conversation directly to GenAI developers. Recent news about bizarre and troubling behavior in ChatGPT that produced “sycophantic” interactions should make us all consider why open dialogue with students about these new tools matters for many reasons, including public safety. In late April of 2025, OpenAI released a system update for ChatGPT that caused the model to engage in behavior that “skewed towards responses that were overly supportive but disingenuous.” To OpenAI’s credit, they quickly rolled back the update, noting “Sycophantic interactions can be uncomfortable, unsettling, and cause distress. We fell short and are working on getting it right”. We cannot trust AI companies, especially as they fight with one another and race (sometimes recklessly) ahead. Part of our challenge is teaching students to be wary consumers and users of this astonishing technology. We want them analyzing and understanding the tsunami, even as they surf the coming waves.

When AI companies release technology to the public and scale it for free, they risk creating a series of second-order consequences that no one can foresee. Our students are heavily using this technology. Some faculty colleagues are lamenting this trend. Other faculty colleagues are embracing this trend. No one among us has any control. We are doing our best, sometimes in contradictory ways, to contend with this imposition from Silicon Valley. Looking ahead, we need better teacher training and coordination. We need more support from our institutional leaders. We need to move quickly beyond facile opinions and misinformed ideals. All of us, no matter our predilections, should talk openly about GenAI with our students. We should teach the debate, engaging the positive possibilities and the worrying risks. Doing so will help our students make the best possible decisions now and in the future, as they choose to use or avoid this technology in their daily digital lives.

The authors of the following issue are doing just this kind of nuanced thinking about Generative AI and education. They are deftly riding the wave. The six essays presented gathered here all grapple with how best to engage students and fellow educators. In "Building Critical AI Literacy: An Approach to Generative AI," Kathryn Conrad and Sean Kamperman advocate for a critical AI literacy approach that examines power dynamics and ethical implications of AI technologies instead of merely teaching students how to use GenAI as a tool. Adrienne Carthon examines the complex relationship between AI technology and HBCU students in “The Stakes Are High. Are the Benefits Bountiful?: HBCU Students, AI, & the Power of Composition.” Kirkwood Adams and Maria G. Baker examine ChatGPT's feedback on first-year writing essays in “Characterizing ChatGPT's Feedback for FYW: Analyzing Feedback Responses to Inquiry-driven Essays.” Kathleen Kennedy and Anuj Gupta outline a thoughtful framework around AI and data acumen learning outcomes in "AI and Data Competencies: Scaffolding holistic AI literacy in Higher Education." In “Enhancing Special Education with Generative AI: Suggestions from K-12 Special Education," Amy Walter chronicles using ChatGPT in a middle school setting. Mila Zhu’s exploration of music generation tools reveals creative possibilities and hidden biases.

These scholars are calling upon us to think creatively and to engage our students fully within this new environment. As they demonstrate, we can teach AI literacy as a fundamental and transdisciplinary skillset. We can do so by fostering spaces for critical examination, by promoting transparent usage, and by creating classroom cultures of curiosity and skeptical inquiry. We can pay attention to the bots, while putting the humans first. In doing so, we will help our students develop beyond blind consumerism, to become informed citizens capable of surfing even the biggest waves.

Individual Manuscripts

• Building Critical AI Literacy: An Approach to Generative AI
  Kathryn Conrad & Sean Kamperman

• Characterizing ChatGPT’s Feedback for FYW:Analyzing Feedback Responses to Inquiry-driven Essays
  Kirkwood Adams & Maria G. Baker

• AI & Data Competencies: Scaffolding holistic AI literacy in Higher Education
  Kathleen Kennedy & Anuj Gupta

• Collaborative Intelligence: Towards Practical, Critical & Cooperative Teaching & Learning with AI
  Amy Walter

• The Stakes Are High. Are the Benefits Bountiful?: HBCU Students, AI, & the Power of Composition Adrienne Carthon

• Algorithmic Foucault: Digital Feminism, the Panopticon, & the Role of AI in Shaping Musical Identities & Pedagogies
  Mila Zhu